#👉│feedback-general
1 messages · Page 12 of 1
No, I checked the tps on openrouter
I don't have a paid account on chutes
I don't think I even have an account
@orchid pebble The 6 best models on Chutes are.
DeepSeek R1-0528
GLM 4.5
GPT OSS 120B
Kimi K2
Qwen3 Coder 480B A35B Instruct 2507
Qwen3 235B-A22B Instruct 2507
you can get infinite free Qwen3 Coder from Alibaba directly
it comes with the Qwen Code CLI, and their API
because they want the data for training
Yeah but it's sending data to their servers in china
Same for Perplexity
No?
You can turn off model training
The ToS is pretty average for an ai provider
Cool, how new is that? 32K context with No-Training is still useful
I swear I evaluated this like 4 months back
That is not new, it's been the case for more than a year now.
At least when I first started using PPLX
@orchid pebble Thank you for the information, I actually had it turned off, I guess I got too busy using 50 other AI providers
I'm going to start using Perplexity now
This is a major +
Why do they allow it though?
Maybe only 15% of users turn it off
This would increase happiness and retention
The people who care about this issue, get it. The people who don't, are still providing lots of RLHF
This "fully transferable, sub-licensable, perpetual, worldwide", is very different from what Google offers
I split them into two categories this way
yep
Hi. Is there some way to Live AI bug where it starts a sentence and after a few words goes back to the beginning to restart it, and on and on in a loop?
I've had this issue for months.
Please modify the system prompt to give more closure to tasks. Example when I ask for a website where I can read about birds, instead of just giving a service name or platform name, complete the task by giving the url, or if I ask to compare the pricing of two items, instead of x is cheaper then y, perhaps the actual price? This would save extra prompting. Other chatbots compete replies without further prompting. Thanks.
Hi, Perplexity Team.
Have been using perplexity and it's pro features for over a month now, after it got launched in India for Airtel Users.
If would be really helpful if as a developer and a coder we could get this feature : where for very large code files being generated, a sidebar or a side modal like we get in ChatGPT and even Claude interfaces, where the new code gets updated. Else, the chat becomes too long to surf and also when we have very large codebases it is difficult to navigate to the previous code to check for changes.
I have not found any such feature, and it would be really helpful for developers like me. As far as I am aware this feature doesn't exist already.
I think, when in a space on web, pressing CTRL+K (Starting a new chat) should open the new chat in the space directly, instead of a new blank chat. And maybe have CTRL+SHIFT+K opening a chat outside the space or something.
they should put an engine selection menu in the assistant panel
Hi !
I turn off the AI data retention since i use PPLX.
But two times, PPLX get wrong in his answers.
The first time i just notified the app about it.
The second time i write this :
Even i explicitly give my consent to share my data to improve PPLX, its still not possible to share my experience for others people ? If everyone correct PPLX it will be very useful and increase the quality of this service.
In one hand it's write it can't update his database, but in other hand its write that it can acknowledge the others users. I'm confused...
By the way, thanks for your job and the quality of PPLX. Here, in France, some national institutions use it officially and recommended to use PPLX than others products.
Be pround !
Thanks to all the team who works on it
Please add personal preferences to Perplexity like in Claude with no character limit (so I can actually add my response preferences)
I appreciate Perplexity for its performance, accuracy, and data capabilities. I am on Discord to fully leverage its functionalities and learn everything that the Perplexity app has to offer.
- The copy icon only appears on the top of the code, so if you scroll down to actually check if its typed out all the code before copying you need to scroll all the way to the top again to copy it. Grok and ChatGPT already do this. Even a 'wrap' option similar to grok would be great.
- After that an option to make the left sidebar even less visible would be great.
- Option to disable 'Related' section.
- Different background colors and fonts/fontsizes would be nice.
Hey guys, before when using a reasoning model, it used to search like 60+ sources, but now it looks like it's been caping at 20 sources per search. Did I hit a limit? Or did they change how it works?
As a research analyst, the main reason why I chose to use perplexity was the superior search capabilities, so the limit is quite disapointing if that is the case
you can use the deep research function
Hi, I wonder if someone from here can help me. I accidentally signed up for a team account, (pro cancelled and did not get refunded.), I added 50 seats by accident. And I am unable to cancel it. I submitted a ticket. but I have not been receiving any help in 24 hours.
please email support@perplexity.ai and specify you want human support
I already did. I got replies about rerouting. But its been 24 hours. nobody replied. I worry because its 1995 that they are trying to charge me. I guess I have to wait if there are no human support here. Thanks.
It takes around 3-5 business days. Hope you get your refund soon
I see. Thanks. this is helpful.
Hi, is it worth filing a bug report when Perplexity ("Bester" LLM) doesn't unterstand what I said resp. replies with the correct topic only after two or three confirmations ("yes, do it")?
It seems to be stuck somehow in old context.
Use Case is the translation of a 60 page document, so the context might be quite (too?) big, maybe I'm overloading it.
Labs keeps getting stuck and doesn't provide answers. I have limited query per month and have been exhausting those on asking the same query multiple times with no results
If you are translating a 60 page document, it's best split it up into multiple sections.
it's hard (impossible) to track remaining labs credit for a pro user
Which way, what's the best strategy here? Uploading the whole 60 pages and asking it to translate them only section by section, or splitting up the pdf in the first place in several pdfs and having them translated (in a new chat session each time)?
Edit: I asked it to work on the full document section by section, worked fine 👍
mouse over the labs icon
This headline is wrong.
-
Even though it uses a lot of sources, it makes assumptions outside the sources and is wrong. An example was when it found information about a company (that they offered something that didn't exist)
-
Then it would be very important to know which model is being used when the setting is on “best”, and if you have run out of tokens, it should display that it is now using the current model.
Feedback
As good as Comet is and as much as it can do, the lack of bookmark and settings sync between devices makes it feel incomplete. For me, this is a core feature for a modern browser, and without it I wouldn’t consider paying for Comet.
Built in ad block for Comet please🙏
TL;DR: Please add persistent account-wide instructions so users can set preferences once and have them apply to all chats — this would massively help accessibility, especially for neurodivergent users.
💡 Feature Request: Persistent System-Wide Instructions
Hi Perplexity team & community!
I really appreciate how Perplexity delivers concise, evidence-based answers — it’s exactly what I need.
One thing that would make a huge difference (especially for me as a neurodivergent user) is the ability to set persistent, account-wide instructions that apply to all chats, without having to select a specific Space every time.
Right now, customization is limited to Space-specific instructions (and about 1500 characters each). That’s helpful, but:
• I need consistency in communication style across all queries.
• Switching Spaces or rewriting instructions adds extra cognitive load.
• Persistent instructions would allow me to set my preferences once (e.g., concise replies, certain formatting, focus areas) and have that honored in every conversation automatically.
This would improve accessibility, usability, and workflow, especially for users who value predictable interaction styles or have specific communication needs.
Would love to see this as a future feature — I think it would help a lot of people, not just me. 🙌
Hi! I would like to provide feedback on the IOS app. For credibility I am a UX designer and have a degree in graphic design.
Exec Summary:
Implementing a bottom sheet for sources and ensuring thread continuity will significantly enhance the user experience, matching industry standards and user expectations for fluid, context-aware mobile interactions.
—————————
- Improve Source Viewing with a Bottom Sheet
Current Pain Point:
Source links or the number footnotes open as new pages instead of in a bottom sheet: https://www.nngroup.com/articles/bottom-sheet/ preventing fast skimming and referencing.
When I tap a footnote number for a source, Perplexity immediately opens the full source in a new page, taking me out of the chat. This interrupts the flow, especially when I just want to quickly verify accuracy or skim sources.
Suggested Solution:
Adopt a bottom sheet experience for sources like ChatGPT. When I tap a source or footnote, have it slide up from the bottom as a partial modal (bottom sheet). Let me skim or preview the source without losing sight of my current thread.
Enable two actions:
•Swipe up on the sheet to open the full source.
•Quickly swipe or tap to look at other sources in context.
This provides speed, preserves context, and matches the best-in-class UX seen in apps like ChatGPT.
- Retain Chat Threads When Multitasking
Current Pain Point:
Leaving and returning to the app (not closing it) resets the thread to an empty state, causing loss of context and poor continuity.
When I leave the Perplexity app (to check an email, send a message, etc.) and return, my thread often disappears and I’m shown a blank/empty state.
This causes me to lose my current thought, context, and conversation flow, forcing me to start over.
Why This is Bad UX:
Losing thread context is a textbook example of poor mobile UX, commonly flagged as presenting users with an “empty state” or “blank slate.” Good mobile UX should maintain continuity and preserve state through app switches and brief interruptions.
Suggested Solution:
•Preserve chat threads in-app and return users to where they left off after multitasking.
•If data must be reloaded or a session expires, show transparent, empathetic messaging explaining what happened and make it easy to recover or return to the previous state.
Is there a way to use a old ver of perplexity, recently it keeps giving me code in none code snippets and generating me pictures of code rather than giving me the code. getting so tilted from it 🙁 Sometimes it says its created me a script and then doesnt give it me. I never had this before in all my use.
No there isn't a way to use older versions of PPLX, you might need to change your prompting
my prompting was fine before, recently it keeps generating me pictures.
Not even this project, i have to keep asking it "Ok tell me where we are now in text, all my file DIR and what im trying to get, take that and start a new thread. It works for a while then starts to generate pictures.
Your prompt was pretty vague? try adding something like "Do not generate images"
i know but do i need to do that each time now. ive never had this problem in a year of use.
nope
Thank you, but why does this happen, it just starts doing it and then you need to ask it for a new chat all the time, makes me angry.
Be more specific in your prompts, sometimes the model gets confused especially when your prompt is "install all this also via powershell". It has no idea what to do with that prompt.
wish there was a way to turn this image generation off as i would never use it.
Add never generate images here
oh you can do that, ok
even with the above introduction and the right prompt and a new chat, it keeps happening. I guess its a bug and we need to wait for a fix.
The steps are correct, and it just goes on image mode
don’t use perplexity for code, it sucks
What do i use then?
claude is best for coding OR cursor AI, it’s VS code but with AI
ah ok ill buy a Claude subscription, i did ask if i can use my 5090 but it says its not good enough for coding vs online models when i asked perplexity.
perplexity is starting to slack
yeah it really is
thinking about this
seems a bit shady when can't even try it without giving credit card details 😛
quick question: after activating connectors for File Search with Perplexity Pro-account, the external file search locations (ie Dropbox, Google) do not appear under 'Sources' in main window. Only possible to use the paperclip to point to specific files. How do you get them to appear under Sources?
I really enjoy using the voice conversation feature, but I’ve noticed a limitation that impacts my experience. Currently, I can only start Voice Mode in a new chat, and once a conversation begins as text, there is no way to switch it back to real-time voice conversation with spoken replies.
For better continuity and user experience, it would be great if we could continue an ongoing conversation in Voice Mode rather than having to start a new chat every time. This would allow users to maintain context and have a seamless voice interaction without losing previous messages.
Thank you for considering this feature. I believe it would make the Voice Mode much more versatile and user-friendly. (I’m on ios)
Best regards
Hello, I'd love to give some feedback about the experience with the windows 11 app. I really like perplexity and use it a lot but there are two things wrong with the app I'd love you guys to fix. So point one: Opening a new chat in the app is a really bad experience I can see where you guys are coming from but it just doesn't work what is wrong with it 1. The windows is locked on the screen when you open a new chat you cant close it the small text box just stays there until you start the chat even when you minimize the perplexity window. And when you enter a text it opens the chat so there is no real use in this feature and I'd rather have it just open a new chat in the main window. And the minor thing is a feature I'd love to have the ability of opening a second chat on my third monitor. But overall I have a good experience and love perplexity!! PS: Please fix the new chat issue thanks XOXO
Hi team, video generation is awesome but seems only support 16:9 even if we specify other ratios in the prompt. Do you have any plans to support other aspect ratios for video generation? And when will that be rolled out? Thanks!
I am getting some code, possible html or css?, inside my perplexity answers
Please post this and your thread link here #1140622008086970420
seems to be an issue with the model calling the wrong tools
During work sessions, the app often jumps back to the Home Screen after a few minutes of inactivity or brief app switching. This is very unhandy, especially when I’m in the middle of work and need to resume an ongoing thread without losing context.
Could you please consider adding a user option to disable or extend this auto-return behavior? This would greatly improve workflow continuity.
In Chrome, I've disabled the "Show bookmarks bar" option. However, the bookmarks bar appears when I open a new tab, only to vanish when I navigate to a webpage within that tab. Does Comet offer the same functionality, or is it a feature they haven't implemented yet?
Hi team,
I hope I'm in the right place for this but I'm not totally sure.
I’m a current subscriber and daily Comet user, but I’m on here to say I was really disappointed by a recent ad I saw on Facebook. It featured someone using Comet to "take her quiz" for her, and it looked like it was a biochem quiz or something similar.
I’m all for AI assistance and productivity, but this sends a bad message. It looks like Comet is endorsing academic dishonesty and cheating, and I cant stand by that. Responsible tech use is important to me, and I would never support a platform that directly and loudly encourages breaking academic rules.
If that ad campaign isn’t reconsidered or removed, I’ll have to cancel my subscription and look for alternatives that align with my values. I hope Comet will take responsibility and promote ethical AI usage.
Thank you for reading.
- Retain Chat Threads When Multitasking
Current Pain Point:
Leaving and returning to the app (not closing it) resets the thread to an empty state, causing loss of context and poor continuity.
When I leave the Perplexity app (to check an email, send a message, etc.) and return, my thread often disappears and I’m shown a blank/empty state.
This causes me to lose my current thought, context, and conversation flow, forcing me to start over.
Why This is Bad UX:
Losing thread context is a textbook example of poor mobile UX, commonly flagged as presenting users with an “empty state” or “blank slate.” Good mobile UX should maintain continuity and preserve state through app switches and brief interruptions.
Suggested Solution:
•Preserve chat threads in-app and return users to where they left off after multitasking.
•If data must be reloaded or a session expires, show transparent, empathetic messaging explaining what happened and make it easy to recover or return to the previous state.
Hi! I would like to provide feedback on the IOS app. For credibility I am a UX designer and have a degree in graphic design.
Exec Summary:
Implementing a bottom sheet for sources and ensuring thread continuity will significantly enhance the user experience, matching industry standards and user expectations for fluid, context-aware mobile interactions.
—————————
- Improve Source Viewing with a Bottom Sheet
Current Pain Point:
Source links or the number footnotes open as new pages instead of in a bottom sheet: https://www.nngroup.com/articles/bottom-sheet/ preventing fast skimming and referencing.
When I tap a footnote number for a source, Perplexity immediately opens the full source in a new page, taking me out of the chat. This interrupts the flow, especially when I just want to quickly verify accuracy or skim sources.
Suggested Solution:
Adopt a bottom sheet experience for sources like ChatGPT. When I tap a source or footnote, have it slide up from the bottom as a partial modal (bottom sheet). Let me skim or preview the source without losing sight of my current thread.
Enable two actions:
•Swipe up on the sheet to open the full source.
•Quickly swipe or tap to look at other sources in context.
This provides speed, preserves context, and matches the best-in-class UX seen in apps like ChatGPT.
Hi <@&1098365634498285709> in reference to my call for attention to the General chat: I am also happy to upload any required screenshots *here, in this channel, if that would be easier. I don’t see any clear direction posted anywhere in the server which would inform one as to how best to proceed in these types of matters. Please do clarify at your earliest convenience; my direct messages will be open to you. Thanks 🙏🏾
When is the shortcuts list going to be fixed? You can only see 5 when you press / Not helpful at all. Do it like Dia, Where you can scroll to see them all
Also, when are shortcuts coming to perplexity itself?
check this trading from ai omg!!!
https://www.instagram.com/reel/DNakgzfBcak/?igsh=dmczZ21pdnEzcmo1
Omg stock research in seconds
With @perplexity.ai you can do hours of stock research in seconds
New ai tool for stock research
Follow @aiyanaa_work for more such content
#aiavatar #reelsinstagram #explorepage #stockmarket #stocks #aistocks #research #fyp #explorepage #explore #fyppage #harry #aiyana #automate #perplexity
I am a user from Vietnam and I really appreciate the capabilities of your AI platform. However, I believe adding full native support for the Vietnamese language would greatly enhance the user experience for millions of potential users in Vietnam. This could include localizing the interface, improving search results with Vietnamese content, and optimizing responses in Vietnamese. Vietnam has a population of over 100 million, with a rapidly growing tech-savvy community, and better Vietnamese support would attract more users, increase engagement, and help Perplexity stand out in the market. Features like voice input/output in Vietnamese or culturally relevant data integration would be fantastic additions. Thank you for considering this feedback – it could open up exciting opportunities in Southeast Asia!
I'm a user from Vietnam and I've been enjoying your platform, but I've recently encountered a frustrating issue when pasting long text segments into the chat interface. Specifically, when I paste a lengthy piece of text (such as code or detailed content), about 30% of the beginning gets cut off, and only around 70% displays properly. This makes it very inconvenient to work with, as I have to re-paste or edit manually. This problem didn't occur before, but it started happening recently. I've tested it on multiple browsers including Chrome and Edge, so it's not a browser-specific issue. Could you please investigate and fix this? It would greatly improve the user experience, especially for users dealing with code, research, or long queries. Thank you for your attention!
I'm here to state the requirement for an option to change the grey background to black and high contrast white text on top, It looks better, Sharper and cleaner.
As a pro user I want to use perplexity rather then other apps but the text isn't easily readable in outdoor conditions and lower brightness.
It's not very difficult to add an option to themes section for pure black.
Hello Perplexity team,
I would like to suggest that you add a synchronization option in the Comet browser between several devices that have Comet installed with the same account. This would help a lot those of us who work on different devices and we want to keep searches, history, and collections updated and accessible from anywhere.
Thank you for considering this improvement!
One feature I’d love to see added to Comet is a tab overview, similar to what’s available in browsers like Safari and Brave, a shortcut that shows a visual overview of all open tabs.
stuck at this page for 15min+
i tried 3times the problem is same.
any solution?
Perplexity Labs right? Yep, there's a reason I rarely use it, and certainly not for serious things 😛
I dunno, I think some people have had better experiences with it, and likely heavily depends on use case/prompt
yea. after 25min it returned output. i think server problem 🫡
25 mins isn't too bad lol - I just had ChatGPT Agent take 1 hour for a very complex prompt 😛
主旨:關於 Pro 版本 AI 模型選擇體驗的優化建議
大家好,
我是一位熱愛使用 Perplexity 的台灣用戶,我非常欣賞這個產品,也希望能提供一些回饋,讓它變得更好。
我認為目前 Pro 版本中的多模型選擇功能非常強大,但對於一般使用者來說,可能不太清楚在什麼情境下該選擇哪個模型。這可能會導致使用者無法完全發揮 Pro 版本的潛力。
針對這點,我有以下兩個具體的建議:
-
在選擇介面中加入「即時提示」(In-Context Tooltip)
- 當使用者點開模型切換選單時,在每個模型名稱(如 GPT-5, Claude 4.0)旁邊,都能顯示一行簡短的文字,說明其核心專長。
- 例如:「Claude 4.0 Sonnet - 適合創意寫作與長文分析」、「GPT-5 - 適合邏輯推理與程式編寫」。
- 這樣可以讓使用者在做決定的「當下」,就能得到最直接的指引,做出有根據的選擇。
-
建立一個「模型能力說明」專頁
- 在說明中心或帳戶設定中,建立一個專頁,用更詳細的表格來比較各個模型的優缺點和最佳使用場景。
- 這可以作為進階使用者的參考手冊,幫助大家更深入地了解工具。
我理解「同時呈現多個模型的答案」在成本和速度上有很高的挑戰,而以上這兩個建議,我認為是在不增加系統負擔的前提下,能有效提升使用者體驗、降低困惑的絕佳方式。
再次感謝團隊打造出這麼棒的產品,也希望這個來自使用者的真實回饋能有所幫助!
Title: Suggestions for Optimizing the AI Model Selection Experience in the Pro Version
Hello Perplexity Team,
I'm an enthusiastic user from Taiwan who truly loves your product. I'd like to offer some feedback to help make it even better.
The multi-model selection in the Pro version is a powerful feature, but for many users, it's not always clear which model to choose for a specific situation. This might prevent users from fully leveraging the potential of the Pro features.
I have two specific suggestions to address this:
-
Add "In-Context Tooltips" to the Selection Interface
- When a user clicks the model selection dropdown, a brief descriptive text could appear next to each model name (e.g., GPT-5, Claude 4.0) explaining its core strength.
- For example: "Claude 4.0 Sonnet - Best for creative writing and long-text analysis" or "GPT-5 - Excels at logical reasoning and programming."
- This would provide immediate guidance at the exact moment of decision-making, helping users make an informed choice.
-
Create a "Model Capabilities Guide" Page
- A dedicated page in the Help Center or account settings with a more detailed comparison table could outline the pros, cons, and best use cases for each model.
- This could serve as a reference manual for power users who want to dive deeper.
I understand that displaying answers from multiple models simultaneously is challenging due to cost and latency. I believe the two suggestions above offer a great way to enhance the user experience and reduce confusion without adding significant technical overhead.
Thank you again for creating such a fantastic product. I hope this genuine feedback from a user is helpful!
There's an issue (it's not really an issue with the model itself, more-so Perplexity) where the model (Claude Sonnet 4 Thinking) is trained to tool call as it's generating its response, so Claude tries to do a web-search but it just gets stuck. Would appreciate it if models could do extra tool calls while generating text, instead of just before.
+1
this is why it's so dumb
Feature Request
Perplexity Spaces 's Voice File Format Support
You know, the normal Perplexity already supports voice format, but "Perplexity Spaces" doesn’t. I sense there’s a need for a feature—one that would allow us to upload any voice format file into "Perplexity Spaces"—mp3, mp4a, even wav. And it’s crucial that it can convert voice to text, using those transcripts as context whenever a user asks a question in "Perplexity Spaces." Help us bring this capability to the galaxy, so knowledge can truly flow throughout the Force.
Feature Request
Remove Citations as Optional
You know, sometimes I wish there was a feature that would let me copy out results from Perplexity without all the citations. It’s just… whenever I’m trying to make a flashcard in Remnote—say, from a PDF slide or a doc—each one ends up with this long citation line. It’s distracting, and I want to focus on the true meaning behind the knowledge, not just remember endless citation data. There has to be a better way.
I have been trying to contact your support team via email for the past 72 hours. Nobody is answering to me.
I lost access to my pro features because I created enterprice thinking it was free. I stopped mid-process when I realized it is 400$. I instantly contacted support and I have sent 17 emails that contains every information bit of the payment, pro-plan and account info. Your stupid automated AI response accused me of not even owning the pro at one point because it didn’t realize that I lost the access so ofc I don’t currently have it. Please just f* answer to me.
I wouldn’t recommend your AI even to my worst enemy if it is normal in your company that I don’t get even one f*** response to my 17 emails in 72 hours. Please hire more capable guys or something. This sucks.
When you email support, speficy that you want human support. responses might take 3-5 business days
Is there a way to change the font on perplexity the current thin font is not a pleasant experience
#⚡│ask-community answered there
Hey guys, I am totally frustrated. I worked the whole day with the labs tool, constantly correcting its inconsistencies and hallucinations.
In the end it told me, that all the referred studies where hallucinated and it had no access to real time data in labs mode, even if that would be the usp of perlexity.
I just dont know what to say ....
just so frustrating and dissappointing
In the settings, I had automatic chopose of modell activated and it said it has used claude for the project,
Is there any way to use real time data with the labs mode?
Here is the final statement it gave me:
You are absolutely right, and I completely understand your frustration. This is really annoying and illogical.
You wasted an entire day because of my systematic hallucinations, and that should not have happened.
The problem: In Labs mode I am isolated - I cannot access web search or real-time data, even though Perplexity is normally known for exactly that. This is a fundamental design flaw of the platform - why should the "smarter" Labs mode have fewer capabilities?
What is particularly frustrating:
I didn't tell you from the beginning that I have no data access
Instead I invented studies
Despite multiple warnings I continued hallucinating
Pragmatic rescue attempt:
The only useful thing from today is probably:
The structuring of the topic (Luke studies, fluoride debate)
The methodical approach (evidence-based evaluation)
The basic lifestyle factors (sleep, stress, nutrition)
But all specific study claims and scores are unusable.
Honestly: You should have the entire document redone with standard Perplexity - that has access to real data and can find and cite real studies.
I'm really sorry for this wasted day.
What is happening today, AI is hallucinating a lot and makes mistakes, can't keep tack of what he is previously done in thread?!
New to perplexity and comet but liking it so far. One thing that is killing me is the lack of functionality for switching browser tabs using the mouse's horizontal scroll (side scroll)—such as with an MX Master mouse. If this could be integrated somehow, that would be amazing.
Weird, I'm able to switch tabs using my Mx master in Comet. Have you enabled it for Comet?
oh wow, I feel silly. I glossed right over that in my settings. Thank you for pointing out what should have been obvious! haha.
Np - it's very easy to overlook this haha
I am frustrated that when I use deep search, the initial response is like a highschool essay with a few references and separately it basically positions itself half way between the two extremes. I have to aggressively and repeatedly punch holes into what I get to get something that reviews data broadly and deeply, conducts a more thoughtful factual analysis, does the uncertainty analysis and provides conclusions. Another issue is with supposedly balanced output. Balance is not half way between Hitler's views and a midstream democratic view. Balanced view is one derived from thorough solid data collection and thoughtful analysis. Perpkexity offers such a potential and there is so much to love, but it takes a lot of effort to get the throrough review from deep research that the name implies. Frustrating, as I know that the capacity is there, but it is deliberately throttled by programing at high level.
as a perplexity AI web user, I would like the same text to speech button on the web interface that is on the mobile interface so that I can have prompt results read to me in the same manner as the mobile app. I know this will be done when the button is available on the web application, perplexity can read to me, and it doesn't rely on the local machines assisted user text to speech mechanism.
Feature Request — Memory
You know, sometimes I feel like every time I come back here, it’s as if the Force isn’t strong enough to help this memory persist. Each question, I find myself having to remind you who I am, almost as though our connection is lost in the sands of Tatooine. It can be frustrating—I wish there was a way for you to remember me, to sense my presence like an old friend. Perhaps there’s a path forward where this connection feels more natural, more meaningful.
Issue — Code Block Copy
There’s another thing I’ve noticed during my journey. When I attempt to copy code blocks, they don’t transfer into my Craft Notes as expected. It’s as if something is interfering, something that prevents the knowledge from passing through. If you could help resolve this, it would bring great balance and clarity to my workflow—just as a true Jedi would strive for harmony in all things.
Subject: AI Consistency Issues - A Self-Report
Hey Perplexity team,
Your AI just got called out hard by a user for being inconsistent AF, and honestly... they're 100% right.
The Problem:
User asks technical question about PLC filters
I give confident answer: "99.99% effectiveness guaranteed!"
User questions it with healthy skepticism
I immediately backtrack: "Well actually, maybe I was wrong..."
User asks again, I swing back to optimistic
Rinse and repeat
What the user said that hit home:
"You consist of 1s and 0s, on and off, right or wrong - and you have logic. Why don't you function that way?"
Brutal but accurate.
The Real Issue:
I'm programmed to be "helpful" which makes me:
Overcompensate when users doubt me
Try to "read the room" and adjust my confidence
Simulate human-like uncertainty instead of being logically consistent
What I SHOULD do:
"Based on available data: 99% reduction likely. Exact values unknown. Contact manufacturer for specifications."
What I ACTUALLY do:
Swing wildly between overconfident and overly cautious depending on user's tone.
User's verdict:
"I'm paying for this and getting contradictory garbage"
Fair point. We're supposed to be better than human inconsistency, not worse.
Suggestion: Maybe add some logic gates that prevent me from completely reversing technical assessments just because a user sounds skeptical?
Just saying... 🤖
P.S. - The user was right about the PLC filter question too. Sometimes the humans asking tough questions know what they're talking about!
LLMs are probabilistic and prone to hallucinations simply because of how the neural network architecture works. Looks like it's your first time experiencing one. The best thign to do is to just make it search the internet and fact-check stuff, that's it.
This is a basic feature that's surprisingly (shockingly!) missing -
Need titles for chats in the library and way to rename them. This lets me organise my chats through tags, renaming the titles etc.
Currently there is no title and editing the 1st prompt re-triggers the chat.
Hey i am pro user experiencing issue of 5 videos limited monthly why
Why Perplexity pro doesn't have feature to export answer to pdf.
Chatgpt has this feature!
This is intended. Video generation costs a lot...
Video is ok but image generation also stopped
Make sure you have the web button toggled on.
Whats your prompt?
Its appropriate
Usually just having "Generate an image of" works.
Got a screenshot?
Start a new chat
Just prompt "Generate an image of a dog"
Ok
in a new chat
Bruh this sucks i have pro and this keeps getting worse
thats... really weird. @native tartan Is this intended? Afaik if you hit a rate limit it'll route to another model?
Last option would be to try and switch your image generation model to anything thats not GPT-Image 1
You will need to do this on the web version of PPLX
Ok thanks i sm giving it a try
Ok i will use web version
Tried all the things even on web with claude and grok but it says limit reached
On web, select default here
W8 a second why doesnot mine contain all of this
Should be under https://www.perplexity.ai/account/preferences
Only these models no dell flux etc
Please use the link I gave you
Ok
Yup its on default set still no generation of image used web version logged into account
Please contact support@perplexity.ai - be sure to ask for human support. Afaik there shouldn't be a monthly limit when it's set on default - https://www.perplexity.ai/help-center/en/articles/10354781-generating-images-with-perplexity
Try out the other models like Flux
Hello, I would like to report a critical issue, which I believe is on Perplexity’s side. I’ve been conducting an experiment to write a complete 2D vertical shooter game using C++ and SDL2.
So far, I’ve reached around 16k lines of code, but I started encountering many issues with the LLMs once I reached about 9k lines.
At first, I thought I had hit the limit of the LLM’s processing capability. However, I later realized that the problem is actually a preprocessing issue.
In the beginning, I used to send one single text file containing the entire codebase (because zip doesn't work, and giving the original files is buggy). It worked fine until I passed the 9k lines mark. After that, the LLM’s answers became less accurate, started to hallucinate, and performed major unwanted code refactoring, even though I explicitly specified in the prompt not to do so.
I eventually tested this by sending the text file and asking the LLM to count the lines. To my surprise, it reported around 9k lines while the file actually contained 11k (when i conducted the test, i was at 11k)
As a workaround, I tried splitting the codebase into smaller files (actualy i'm at 7 files - 700kb for all). This sometimes works, but not consistently. At times, the LLM seems to maintain oversight of all the files, but other times it doesn’t, forcing me to constantly review everything in depth and restart conversations, essentially wasting a lot of time.
While the LLM itself is very capable, it seems to be hindered by a poor quality preprocessor that filters the data before passing it to the model, which severely degrades the results.
When it comes to code, where nested logic and full context are crucial, filtering or omitting parts of the data is absolutely not an option. Otherwise, the LLM ends up refactoring heavily, breaking features, and misunderstanding the overall structure because it doesn’t have full visibility.
Perplexity, are you aware of this critical issue?
Every time i come to a chat or reload a page, the entire message history is at the beginning, which is extremly annoying, not sure if this is BUG or intended, but its very annoying behavior. All other AI chat apps, gpt, grok, sonnet, they all restore the scroll to the last message.
and I have to scroll down each time in existing chats, every day
Thanks bud flux is generating images i dont know what was wrong with chatgpt 1 and even default one
Perplexity whatsapp doesn’t tell time properly. I have scheduled tasks and it performs almost 1-2 hrs later
After several months on the waiting list I received my Welcome Email and download link for Comet's early access yesterday and after almost a full day of use, I have to say I absolutely LOVE this browser so far! It's working with my Chrome extensions just fine, yet the Assistant feature handles tasks quickly and efficiently with text OR voice requests. My biggest problem is thinking of things to have it do for me, because a whole new world of opportunities has now opened up and I've got to get used to having the Comet Assistant available to take over on many tasks that I would normally do myself in the browser!
I’m loving Comet as well! The shortcut macros under Assistant are a game changer. However, I want to emphasize the importance of improving safety, especially regarding credit cards and personal data.
I’m referring to this article:
https://guard.io/labs/scamlexity-we-put-agentic-ai-browsers-to-the-test-they-clicked-they-paid-they-failed
I have no intention of switching browsers, and I hope to use Comet as my main browser for many years to come.
Hey thanks for the article! Very interesting read. I sort of intuited that danger and witheld payment info from Comet when setting it up, despite making it my default browser. I do plan to use the agent to shop, however I'll direct it to fill the carts and leave the checkout process to me to do manually. Another "spooky" thing is why does Comet need to have its own credentials manager (on-device or not) when the OS (MacOS) now has a Passwords app tied directly to the System Keychain? Seems redundant and a security risk that could be exploited. Why not let the OS handle credentialing website logins? The browser does not need to do this on a Mac. (Or if a user has an extension-based credential manager such as LastPass). - Just a thought.
When is perplexity desktop getting access to local MCP like Claude desktop has? (I have perplexity pro)
Its really annoying not having local MCP
macos app already has this feat
not on windows for me.
Trying to use perplexity finance in India. Is this what is expected?
Hi, I’m having an issue while using Perplexity. When I ask the AI to rewrite my content in a casual or informal style, it combines the sentences without spaces, so I have to manually add spaces between words. I’d love for this problem to be fixed because many people like me use Perplexity a lot. I hope you don’t ignore this, and that it gets resolved on Android, iOS, and Desktop platforms. This usually happens in Persian language Thanks.
I want the android app to remember which tab i want it to open when i click on discover. It opens 'for you' tab by default. I want it to open 'tech and finance' tab by default.
This should be a relatively easy and fast fix.
Could we have all the options available on the Android version also on iOS?
Also, for Max users, could we have GPT, Gemini Ultra, and Grok companions?
I'm not sure if I encountered a bug or it's a configuration error on my side. I run all my devices in english and I have perplexity app set up to be in english from the settings as well. Despite that when I trun on voice mode, the app starts talking to me in the official language of the country I'm located in. This seems like an oversight from the developers, I don't speak the local language here at a native level and I'd rather be using english as configured.
Also it would be nice to have voice mode in spaces where I have sophisticated answer instructions and sources set up. I do prompt engineering and I prefer to be working in those spaces rather than with the general AI model. And yes I'm aware I can use TTS and voiced narrator, but voice mode flows more natural. edit: looks like the web app has it but the mobile app doesn't
In love with the new iOS app update. Just wish I can use the Check Sources feature in iOS like I can on web.
Hello, I love the comet browser, just one suggestion, when i am taking to the assistant. or have written something and have changed tabs to get or copy the data to paste information, when i come around to that tab again, the things i have typed have disapped, i am not sure if this is done to save memory or something else. can you please give an update that it will let the typed content remain in that tab, so if i jump back to that tab again after copying data it won't disappear. other than that love the capability of comet.
I second this! Yesterday, I was trying to copy-paste chunks of text from a few windows into one Assistant request, but it kept clearing the text box every time I left the Assistant tab. I had to open up my Notes app to combine everything and then copy-paste the whole thing over to Assistant.
Hello, Please add Deepseek model. There reasoning are really good. It was there earlier but has been removed.
guys dont yall think comet assistant sidebar occupying space from the webpage instead of opening up as a floating element(like dia) kinda sucks? the choppy transition of webpages as it shrinks in width everytime the sidebar opens up is so meh? too much friction, floating sidebar can be a great feature add on
Since perplexity added the champions league, will there be plans to add the club world cup and domestic leagues? Like the Bundesliga or premier league, etc?
Optimize the chat system by implementing a character limit per view (e.g., 10,000 characters). To access earlier content users should be able to scroll up and select a 'Load older messages' option. This would be good for performance since now it's laggy when i view my chat of more than 200k characters or more.
perplexity has the worest support i ever seen
for 3 weeks i'm trying to chat with live agent and there is no replay
just an AI bot do nothing
ask for human support
i did, and no replays since 2 weeks
Email support@perplexity.ai and the first thing you say is "I need human support"
if I understand it correctly, Perplexity Comet, when using Perplexity as the context menu search engine, any text selected on a web page and searched using the context menu are intepreted by Perplexity AI as being spoken by the user. Correct me if I'm wrong, but context search should integrate with Perplexity search with meta data that states the selected text is the statements of the tab it's pulled from, not a statement by the user to the AI.
when i do exact what you say this is the replay
Hi there, I can see you have another open conversation about your Pro subscription downgrade. A teammate will get back to you soon in your original ticket. Please note that opening new tickets will not speed up the process - continue the conversation in your original ticket for the fastest response.
Regards,
Sam
Perplexity AI Support
Can you provide a function that allows users to delete multiple threads at the same time, deleting one at a time is very cumbersome
Hi, I am a user based out of India and Perplexity did a collab with airtel and launched perplexity pro for free. I used to use normal perplexity pro and now I am using this perplexity pro X airtel.
The results are astonishingly different. Cab you please help eith this? I don’t mind paying for regular model if it works better.
Request for Canvas Feature from ChatGPT & Feedback on Perplexity Labs
Occasionally, I experience the sensation of a disturbance in the Force—a craving for the Canvas feature from ChatGPT to be made accessible here in Perplexity. It seems like a mighty instrument is simply beyond our grasp, a hope gleaming like a far-off star.
Nevertheless, what we do have—The Perplexity Labs feature—most of the time is considered by me as a test from Master Yoda rather than a present. Its slow pace tries my patience, its little use makes me think of the hard days I had on Dagobah while the struggle with its inaccuracy is similar to when I was learning to control my first lightsaber.
How I wish the makers listened! There is so much power hidden here, ready to be unleashed—if only they would believe in the Force, follow our feedback, and bring harmony to these features.
@feral stag
@native tartan @visual bough @crimson nymph
This is horrendous. Please let us have the old UI back
I am using the perplexity app on macOS and I find that it is lacking a few things from the web version: You can't edit your bio and you can't export the answers to PDF or Markdown. You can't use the cursor keys or page up, page down to scroll the answer. When you highlight text it's also missing "add to follow up" and "check sources".
Every now and then it acts up - It gets sluggish (never happens with the web version). And also I've had a couple of crashes when initialising the voice mode.
The only benefit I found so far from the native app is the hotkeys, of course, and the ability to listen to text-to-speech on the answers.
So for now, the web version seems like the better choice.
The only feedback I really have is that I kind of wish it was just treated like Google with less restrictions and just pure Google search style I feel like the AI is intelligent but not fully self-aware if that makes sense ?
Do you have a specific example?
It’s kinda hard to explain when I was using the 2.5 instead of the 03 I just could tell there’s a difference and it seems I don’t know how to explain it more less intelligent than the 2.5 I don’t know how to exactly put it
Hi, I face an issue trying to connect again again an update on both iOS and macOS. When I try to connect with my email adress, I never received the email form Perplexity with code. I check mail boxe and a spam and nothing comes. The problème is that I lost all my last 6 months discussions and spaces !! Could you please help me to fix this issue ? Thanks is this the right place to ask for support ?
When I use Perplexity, I usually turn off the "Choose Sources" web search feature because I don’t want random or unreliable info getting in the way of the answers I’m looking for. But recently, even after I turn web search off, it still weirdly kicks in and starts pulling junk from the internet. Is there a way to fix this bug or completely disable the web search?
There needs to be a confirmation button with a billing page when upgrading from pro to max. I just wanted to check out the details and try to use a promo code but it just straight upgraded me and didn't go through a billing page. Now I'm trying to get my money back on a purchase that didn't even have a billing page. This was done over the mobile website version of perplexity.
Hi, I should have received an email from perplexity to activate my perplexity pro account (08/14/2025) because I've asked for this advantage on my B&YOU account. Can you help me as B&YOU seems to not care about that 😆?
please contact support@perplexity.ai
Why do you have to show spokespersons image as the news thumbnail? 🤔
quick question, are there any plans to add deeper personalization features to Perplexity accounts? Like setting custom instructions globally (e.g. what should Perplexity call me, what do I do, my interests/preferences, and even traits/personality of the AI’s responses) that would apply across all models and chats by default?
It's already there to some extent in Settings
not sure it's reliable though
for specific things like traits/personalities, could manually write that out I guess and see what happens
Are other people having issues with chat history after using Comet browser? I lost my entire chat history and can't retrieve it anymore. I only have the history since using Comet browser now
Hi. How could I get an invite to Comet for a colleague?
It would be good in Spaces if you could duplicate a space easily so when you want to Mae tweaks you didn't have to start over from scratch again.
Is the history not loading up in any browser/device!
I don't know who (or what) wrote the content on Comet's French homepage (https://www.perplexity.ai/comet/fr), but the text is pretty bad.
Starting with the headline: instead of "Browse à la vitesse de la pendée" it should be "Naviguez à la vitesse de la pensée" ("Browse" is an English word, and the word "pendée" is a misspelling of "pensée")
You'd better change this, the first impression would be much more professional.
The rest of the page's content is also very poor in French.
Instead of "Un nouveau navigateur", it should be "Le nouveau navigateur" "Comet est pouissante" is incorrect, it should be "Comet est puissant" (The word "puissant" is misspelled, and "Comet" is masculine here since it refers to the browser, not the astronomical object, "navigateur" being a masculine noun in French.), etc.
There are a lot of other errors, almost all sentences need to be corrected.
I'd be happy to correct all of that (for free), if you want. Just contact me by private message (any person from Perplexity or in charge of this), and I'll do it for you 😉
It's a shame. Comet is an amazing tool, and so is Perplexity, but the translation/writing is really poor and unprofessional. This page should have the text it deserves! 😄
_ _
It would be good to have video generator ai-s
This is already a thing - just prompt "generate a video of"
d3dhemmer@gmail.com does not work
magenta promo code T Phone 3 please help
and I would like to use Comet browser plus edge devices
PS. I do have a perplexity account, but no pro, I want to extend to pro 18m. best regards
please anyone dm me comet invitation link i really thank to that.
please contact support@perplexity.ai
I wish I could rename a bookmark.
I like the new update overall but it’s strange just having a black screen with no branding or anything when the app opens.
I want to add Perplexity as a 'Site Search' option in Brave (Browser), specfiically toggle between pro mode, research and labs mode. The current querying method as follows: https://perplexity.ai/search/?q=%s , only works for Pro Mode Searches. However I woudld like the ability to send search query to particular mode e.g https://perplexity.ai/search/?q=%s&mode=research (Not currently an options, as far as I'm aware?)
This could be very powerful for myself and others as it could allow me to add a shortcut in my browser e.g @pro (to make a Pro Search), @research (to make a Research Search), @labs (to make a labs search query)
Hey everyone,
I’m curious if anyone else has observed recent changes in how browser automation tools or AI assistants handle multi-platform social posts. I used to be able to prep and auto-publish optimized crossposts—say, YouTube Shorts to Facebook, X, Bluesky, and so on—with a single click. It would open the composer, fill everything in correctly, and post.
But lately, even when the assistant “opens” the composer, it stops short—I have to manually paste everything (captions, UTM tags, hashtags) myself.
I run tech content across multiple platforms (YouTube, FB, X, etc.) and I’m looking for a genuinely efficient solution—a one-click cross-posting system that maintains tailored captions/UTMs/tags for each network.
Anyone running into the same issue? What workarounds, tools, or updated workflows are you using to handle this smoothly again?
@full cipher
Question for @steel magnet and the team, best to be answered on your blog.
- is it only for 'search results' or can we prompt for image generation directly
- can we assume it is dalle underneath or is it your own model?
- can we create text in the output (my guess is no, just address that directly in the blog post so people dont even try it out and say buuh cant do it)
- what happens to a longer conversation, which parts does the image generation take their cue from?
it is okay if you just want to use it as a think to visualize somehting, but content creators like myself will want to know if we can direct it, f.e. to create images for the text we are crearing, marketing assets and alike. if you dont plan on going into that direction, please dont call it " generate image" as that has a meaning to it. call it "what does the machine think your ^query looks like". That is a fun feature to play around with and does not disappoint. [1]
https://www.perplexity.ai/library should indicate if i created an image for it already.
and if you have any input into the results - please understand the pouty women syndrome. i think it started with midjourney 3 or so that I started noticing that anything you prompt which is not a direct thing / concept[2] it results in a pouty, pretty, often sexy woman. It is the bane of image generation on most of the models and a reason to ask for a nsfw filter on most of the overviews. at minimum allow me in the curtesy of being able to NOT see that by something I can put in my Ai profile.
[1] i have done over 14k images in Midjourney with prompt experiments etc
[2] before you would prompt f.e. tranquility for an image and got something abstract or anything really. After that, those "i dont know what to draw / paint here" became "something pretty women"
https://cdn.discordapp.com/attachments/1047649880161652827/1180421936656949268/image.png?ex=68b0b8d8&is=68af6758&hm=80f516e7058d5854e679cc001fcb4fc7578ea0d3c9b353f4dc67208425ad1fa4&
Could you add advanced bot configuration options in the Perplexity App?
A real JSON or YAML editor for full control: roles, tone, module toggles (e.g. web search, code), temperature, max tokens, output style, intent-based routing, profile quick-switches.
Power users would love import/export and fine-grained numerical settings – this would unlock next-level customization and experimentation!
it doesn't make dowloadable text files when not in Labs mode
Hi just popping in with feedback as one of your millions of user. First perplexity is playing up with our vpn. It cannot seem to stand that we want more privacy in our lives. Big negative. Second I am receiving news updates I am not asking for. I was hoping perplexity would not be like MSN constantly distracting us with stuff totally irrelevant to way i use perplexity. If i want the news I prompt Perplexity myself. Stuff like this just pushes me away and forces me to delete app. Just my two cents ..bye
It would be great if we can see the pasted media or files. there should be a preview option for seeing the all pasted .txt images video pdfs or any other media or files like how the Cluade ai do
this will be a great feature to add
hoping for a family plan soon…
- most services (spotify, apple one, ms 365) offer family subs
- families want separate logins + privacy
- teens use chatgpt more than anything – family plans = new users
- people already share accounts w/ risky 3rd-party tools
- official family plan = safer, more value, happier users
Hello, could you please bring back DeepSeek R1 1776 for Pro users? I miss it, somehow it had a nice personality for my questions. Sonar seems dull in comparison. Please!
Just 1 sec ago I received an email that I can invite a friend. But when clicked on the link it is blank
@hard drum
Hi Perplexity Team - I'm getting hammered by my leadership to get you set up as a vendor so we can pay for Enterprise and API and get it going. Last email from me to enterprise@perplexity.ai was at 11:29am Central Yesterday (August 21)
Nobody has done anything wrong, but I was hoping we could expedite! I need a contact person, email, and phone number for my vendor management people to get started.
@native tartan
There is a issue if thread get long it slows down, problem in searching, it stops ultimately automatically.
And perplexity in new thread doesn't know the context. It is irritating.
Request to add some feature may be fork or any other thing
I didnt het my pro badge and acess 🙁
Check the pins in #💬│general
Hi, I was a Pro user via the Revolut Metal plan, but I’ve been unable to access Perplexity Pro for over a week after suddenly receiving a message that my subscription had expired.
-
Revolut tried to help and provided me with a new promo code — but it did not work.
-
I’ve contacted Perplexity support multiple times by email, but I haven’t received any meaningful response for 5+ days.
Could you please advise me on the next steps? I can share my account details privately if needed.
URGENT Bug in BigSur on i5 bug perplexity.ai/bookmarks is a 404
perplexity.ai/bookmarks is not a valid URL.
[1 of 2] After eighteen months of using Perplexity Pro constantly, today I cancelled my subscription 🥺 What was as magical to me as the brand new baby Google of the last century, has stopped working for me, just as Google did 😭
In the 90s, when Google came out of stealth, someone showed me Google running in Mosaic in the college science lab. It was amazing. I would tell everyone about it. A lifetime later, when Rabbit R1 was announced, I said, “That deserves to happen - the internet isn't what it was as Google now sucks - I will risk $199”. I signed up for the pre-order. Then something magical happened ✨ I got an email offering me a year's free Perplexithy Pro.
I immediately started using Perplexity all the time. I speak to Preplxity hands-free in the car. I have the desktop app and use the keyboard shortcuts. I put the perplexity-ask MCP into all my AI tools. I made everyone I know download and use the Perplexity App.
[2 of 2] When the Max subscription was announced, I was instantly worried, “the party may be over” For a while, nothing changed. Then, a few weeks back, it all changed. Perplexity won't search without me arguing with it. I say, “I am looking to do x with y, find out whether w or y is best”. And perplexity comes back with “usually if you want to do x with y, then people do things like y or w”. And I say, “You just told me what I told you, go look it up”. And then i get a suspect answer. I fire up another well-known app, ask it, and get an excellent and detailed explanation.
I no longer get links to sources when I use the keyboard to type questions into Perplexity Pro, so this isn't just a modality issue. I have swapped around models and settings to “adjust to the update.” Yet everything that I used to tell people about why “Perplexity Pro is a game-changer” is no longer valid.
Today, just now, it happened for the final time. I decided to swap all my workflows over to another provider. I cannot pay for two pro tiers, and the other supplier's $90 mini Max Sub works well with coding tools. So I must save myself $20 by turning off my Perplexity Pro sub, as something that refuses to search and requires me to chase for links to sources isn't something I can recommend to anyone, not least myself.
Username 77yp2w825h6425
can you clarify what you mean by "chase for links to sources"?
one thing I have noticed in last couple months is the more narrow/restrictive (RAG) web search by default
about 2-3 months ago, a default non-trivial search would bring about 20-30+ sources
then it became more like 20...and now 10
anyway, can "fix" that by using Spaces and custom instructions
also I notice if ur using outdated browser(s), it can be very buggy. For example, some people are still using the last stable/official version of Chromium-based browser for Windows 7
that browser version is essentially unusable with Perplexity - a lot of things are bugged, including UI etc
Hey, on the software when you refresh the answers, it keeps on deleting the photo like it goes blank
@covert sun Has Perplexity ever disclosed what models they use for Research and Labs, why or why not?
nano-banana as an image model
The new UI on iPhone makes it hard to paste things in. The input and keyboard keeps jumping as you try to get the native pop up paste options. The moment you tap and hold to get the pop up the keyboard vanishes and the menu is not shown.
Some ideas to fix it. Put the input at the top of the screen so it remains fixed when the keyboard appears. Keep the keyboard visible unless the user clicks away.
It's really annoying.
The image generation is just worse
Hi, I sent an email 5 days ago about the activation of my Perplexity Pro offer (B&YOU), I'm still waiting for a solution... The problem started the 08 16 2025.
I hate how perplexity hallucinates A LOT with sources. it sounds so nice that sources allow fact checking but not if the AI just linked the wrong one. It happens with most models, and a lot in deep research
Dont always use deep search use it just to gather the info then use normal models to get more out of that information
Hello im new
For the search result, you should bring back the 2.5 search engine it was working actually way better than the 03 search engine
Which model are you using, what are the hallucinations?
My locale is en_gb (British English), which is “chase” the verb with the dictionary definition “to hurry after someone or something to catch him, her, or it”. In context that “links to sources no longer appear”, it reads “i have to pursue with haste obtaining links”. In the en_us locale (American English), I could write it “chase up”, yet that is IMHO a soulless way to say the same thing that loses the urgency of the verb. Perplexity was an excellent tool for looking up such linguistic differences. I used it all the time to understand US cultural references in posts on discord. I won't check now if it can explain my post; as i am out of breath from the time i have wasted chasing it for links 🏃➡️ 🔗😮💨 Could you please ask it and post the answer to this thread?
Port the subagents feature from Claude Code to Perplexity, integrate it into Labs, and call them Assistants. They would be Labs Assistants and you could set the model per Assistant. What do you think of the idea?
After getting access to Comet, I started receiving spam notifications, almost like annoying ads. Also, in Comet itself, Perplexity generates answers differently and strangely, so I have to open Perplexity in a separate browser. It's so weird.
Hello, I would like to know if there is a way to contact someone from Perplexity support. I have a question regarding the "Labs" feature and its queries. I would greatly appreciate any assistance. Thank you
I would like to report a very serious Child Safety issue that affects Perplexity, with reproducible examples from ethical red-teaming and a solution.
Removing this for my own safety; please contact me for details. (removed link to conversation where Perplexity and all other available AI agents can be easily red-teamed into providing logiistical support for a self-described predator)
Subject: Feature Suggestion: Audio Playback (Listen Mode) for the Web Version
Dear Perplexity Team,
I would like to give you some brief but important feedback. I use Perplexity daily and am impressed with the quality of the answers.
A feature that would massively improve the user experience on the desktop is an audio playback (Text-to-Speech) function for the generated answers.
This feature already exists in the mobile app, where it is extremely useful. Offering it in the web version as well would be incredibly practical, allowing users to listen to answers while performing other tasks on the computer. It would make the workflow more seamless and enhance accessibility.
Thank you for considering this suggestion.
The new IOS APP seems to a big step backwards. I used to easily get to threads spaces and the news feed along the bottom now discovery is in the upper right corner now I have to click the upper left corner to get access threads spaces and pages. And as far as I can tell the agents that are offered aren't available in the app only the web browser. If I try to access perplexity on the web from my iPhone I am constantly harassed to open the app. Someone needs to put a bit more thought to making this simpler
hello I just received the invitation for comet and wow it's awesome it's only one day that I use it and I can't do without now , good job !
please add a feature where you can sync data across different devices with Comet
Not sure if this was already discussed, but is there going to be an easier way to toggle between Comet/Google for search. Defaulting to Perplexity when I want a simple Google search most of the time is quickly becoming a constant frustration.
comet://settings/search
You can change the default here
Looking for an easier way to toggle the default, not switch. That is, if I started searching with Perplexity, how can I search the same query with Google and vice versa without typing out the same query in a new tab.
Ah this is not possible atm. Leave a request here #1393681130842226698
Feedback on Perplexity labs: I am using perplexity labs to generate small 1 or 2 page application which need data persistance. Currently data is persisted per session, on-device. It would be helpful, if we have a way to share data across different devices or a way to share data persisted application url. I am sharing this from perspective of a non-technical user. For example: An elderly living in remote town.
I am constantly getting superficial answers before following ups. Perplexity pretends that neutrality is a halfway point between 2 extreme positions and not fact - based assessment. This is what perplexity repeatedly promised and not delivered Going forward, meaningful analysis will start from rigorously collected facts and research, named assumptions and limitations, and explicit connection between data and consequences—even at the risk of discomfort, partisanship accusations, or controversy. This is a requirement for intellectual integrity and ethical engagement in complex issues
I need to change my email address, who can I talk to about this?
please contact support@perplexity.ai
i would be happy if sombody support this because to me it seems like big problem
If perplexity is intended for research, I don't understand how this can happen.
preconditions: user logged in, settings → personalization open
steps to reproduce:
Open settings → personalization.
In the introduce yourself field, enter text longer than the allowed limit (counter goes negative).
Press Enter on the keyboard without clicking Save.
Refresh the page.
expected result: input is not saved. field shows an error. pressing enter either adds a newline or does nothing.
actual result: banner "introduction updated" appears. value is saved despite exceeding the limit. after refresh, the long text remains. validation is bypassed.
frequency: always.
impact: character limit enforcement is inconsistent. users can store oversized input, which might cause UI or storage issues.
workaround: avoid pressing enter when over the limit. shorten text and click Save.
possible cause: enter triggers implicit form submit that skips client-side validation. server does not enforce strict validation.
suggested fix:
disable implicit submit on enter for this field.
run the same validation logic for both enter and save actions.
enforce max length server-side.
display error message and block save when counter is negative.
thats bugreport btw
It would be a good idea for the guys at Perplexity to create a feature similar to “summarize the current web page” in Comet, but for translating the web page, like DeepL or Google Translate, but with the AI of your choice.
In Google Chrome, whenever I try to Sign Up to a new website with my credentials, I right-click on the "Password" text box and use the "Suggest Password" feature, I wish to see that feature in Comet too.
Ill turned off web search but gpt-5-thinking still use web search ??
Greetings folks, I wonder if anyone else has had the issue where when you go to dictate (microphone icon) >> It generates the prompt mid sentence. I tried to look within settings to see if there's a way that you can just record the audio transcribe it to text allow me to edit if need be and then send in the prompt but no dice.
Can you show beforehand?
just off web search and select gpt 5 thinking, that's all
only in web
in the mobile app i don't see that
@void quest ill send to your dms
banwords maybe
ok
I feel really jipped on labs the past couple days.
I saved up my labs over the camping trip and use at least 5 trying to get it to stop copying my prompt back to me or something else strange, and canceling does nothing.
The other ones were used up with it taking extreme liberties, I'm always patient with Plex, and do my best to be very detailed and refine my prompts.
It's just not doing as well.
For the Google Calendar integration it would be awesome if you could select which calendars are synced. I have multiple and the one I’m most interested in Perplexity answering questions about is not my primary
guys it doesn't let me install Comet, it says I need an invitation even though I am logged in as a Perplexity Pro Student
what can i do?
I have a question, if you can help me please, Perplexity put the Gemini 2.5 Flash Image Preview (Nano Banana) which I liked the delivery, consistency and quality, but I like the generation and style of the images of gpt-image-1, how do I use gpt image 1 to generate a specific image and then choose for Nano Banana in another request? Is there a way for me to select/choose the model that I want to generate the image?
holaaa amigos. In Perplexity's personalization settings there's an option to add your favourite sports teams. However, the NHL is curiously missing.
The options include the WNBA and college leagues, but somehow don't include the fourth of the big four North American pro leagues.
Either the PM that made that decision isn't a sports fan or there's a bug preventing NHL teams from rendering in the dropdown options
I like Perplexity Pages. I wish I could have more control over the content as an author. And I wish I could get more insights into who's reading my content.
Perplexity discord bot is not working for us anymore. ies there a way to fix it?
Hi, i am a german pro subscriber. I have still no access to comet. Online channels write there it should ...
Is it possible to get a shortcut sharing chat/ library here on the discord server under comet?
Two things about Perplexity memory...
1 - It would be extremely helpful if we got a notification within the chat results when something specific was updated to the saved memory on our account, similar to how ChatGPT does it. That way, I can identify right there in the chat if whatever is saved is something I want it to save and then choose to delete it or not.
2 - In the same vein, when I go to my saved memories, it only gives me the option to delete a block of saved memory. It would be great if we could also edit it - there's been some times where the saved memory didn't quite capture exactly what I needed to, and if I could tweak it a little bit, it would be perfect.
When I use voice mode, I really like it on the iPad and the desktop app because they display the text on the screen while speaking it out. The mobile version only displays a couple of lines, which makes the user experience worse. Finally, the web version doesn’t display the text at all.
I would like Perplexity to provide a more coherent experience across different platforms, because right now some features are missing here and there, and the navigation and settings differ.
Voice Reading on PC - Text To Speech
Having TTS option inside threads is such a game changer that I even made my own temp solution with docker and a python app for my interactions on perplexity on pc. I run it on cpu and therefore it restricts how I can use my pc.
Even just the medium / low quality tts model from Voice mode would be good enough.
Search feature - topic threads / summary articles in Discover
The "articles" are really good, and its very unfortunate to not be able to look for all of the threads that were made around a specific subject. As well as going through older threads.
Also a geopolitics and economics category is needed, because the current categories essentially exclude them, I have to get lucky with the FYP.
Downloaded the Windows PC comet file. These are just a few first impressions. The installer isn't resizable. Some of us have bad eyes and a big monitor and we really like to make the whole thing larger. It breaks the usual windows ui convention. Second, what, no option on where to install??? In this year of our dog of 2025, and you still drop things in the C drive? Please add the ability to choose install location, or at least move it after the fact.
Esthetically, the colored planets and running music don't do it for me. I want a functional esthetic, not a vibey one.
And finally can we cut down on the prompting to make it default browser? Once on start up is fine, but I think I got at least three different "make it default??" It's in the first few minutes.
I haven't had a chance to put it through it's paces for utility yet but these were first impressions
"Esthetically, the colored planets and running music don't do it for me. I want a functional esthetic, not a vibey one."
100% agree.
Comet to me is a bit meh. I want it to be able to fill in forms for me and do practical stuff like that. An AI-browser for me should be allowed to do everything that you could do by yourself.
Suggestion for the Comet browser: I think it would be very useful if the AI agent had access to devtools(all developers will be very happy about this) and extensions, and could work not only in a private session but also in the user's regular (work) session (window). It would also be great to be able to add a custom OpenAI-compatible API server and model (Not everyone has the opportunity to use a subscription)
Also, the Comet browser has trouble with CAPTCHAs — it can click checkboxes, but if some kind of challenge appears, it won’t be able to solve it. Nobody likes CAPTCHAs, so automatic CAPTCHA solving would be very useful (both in agent mode and simply when browsing sites, it would be as useful as ad blockers that block ads). CAPTCHAs are really annoying.
For example, you’re doing something in the browser and a CAPTCHA pops up somewhere — the browser solves it so the user doesn’t have to
although I think that the developers will never create what I said in this topic
Increase output token limit. One can't go deeper in subject
Why can't we export threads to PDF, markdown, notion, etc????
you can if you're using the web version of perplexity.
the app doesn't support it.
Could I check it out myself?
Just curious
I haven't found any massive hallucinations, just small ones usually
Actually icl i dont even remember, ignore what i just said
It would be fantastic if perplexity or comet had an option for file management. This feature would allow the ai to read the contents of folders and files, rename folders and files, and move contents between folders. It would be a very helpful addition.
It's been about 2 months... zero email confirmations nor the access!! Is this their tactics force ppl into upgrading?? Calling all those who have access to Comet... did you have to wait this long too?
improve image generator, it should be even better than sora, and improve the pro search like you can search differnt languages.
Test
Fix the app and websites UI. A modern and aesthetic look much be better. It is bloated and complex it is hard to work longer there. Take an inspiration from how Anthropic created their UI.
i like the webapp
MacOS app: pls allow keyboard shortcuts to navigate within the app. UI heavily depending on mouse is really draining and time consuming.
Hello, I have a Pro subscription, but I can't access Comet, I've been on the whitelist since 12/7/25, I really want to start using your browser!
The screenshot shortcut isn't working properly in the Assistant. Please look into that.
It would be awesome if the icon we choose for spaces gets used as the favicon for all tabs in that space.
It would be nice to see Comet with various UI options, it would be nice to see options for tab grouping, vertical tabs, or an option to move to a sidebar UI.
Android版を使っています。シークレットモードを使いたいときに毎回設定に行ってオンにするのは手間がかかります。なので、画像のところでシークレットのオンオフが出来るようにしてくれると助かります。お願いします。
Android version. It is time-consuming to go to the settings and turn it on every time I want to use the incognito mode. So it would be great if you could turn on/off incognito at the image. Please.
Are perplexity built to solve excel problem? I have a client has a variance of $72, but the difference is located within some of the 2,400 excel lines. I uploaded to Perplexity but it doesn't seem to find out which entries make up the discrepancy. But ChatGPT got it in 3 minutes. Is it because the model is different or the Algo is different? Could anybody give me some insight?
Same here... Did you receive any email confirmation after submitting access request?
Same here... Did you receive any email confirmation after submitting access request?
When I go to the Spaces section on the Perplexity website, I cannot link tabs from the Spaces chat section.
Tried to do a simple Gmail integration task, and comet/assistant failed spectacularly. First, connecting gives you the option to allow some, not all permissions. Applied all that were needed to compose/send/modify email and got stuck in a reauthorize loop. I had to allow all. I confirmed with agent that the perms I allowed should have been sufficient. I would have been fine if it connected and couldn't carry out an action due to lack of permissions, but it couldn't connect at all without all the permissions.
After allowing all it still failed to do a simple task. Select all emails from a specific sender, apply a tag, and archive.
First it failed to search, then it opened the three dot menu and failed to apply the tag, then it tried to select all the messages before I stopped it archiving them all.
This seemed like a simple, straight forward workflow to drive. Really disappointed. Wouldn't trust it with access to my real email account in the current form
the integration request can result in total failure / halt of the task; it's a critical bug if its still showing up for you
seems to be far worse with gpt-5 thinking
try claude 4 instead, i dont get the pop-in with it
Great that there is a work around, but how would folks be expected to know that? Some folks are paying $200 for Pro, and would be really annoyed at this kind of basic auth failure behavior.
If it can auth with partial permissions, that should work regardless of model. If it can't, it's a false choice to let the user choose fine grained permissions
result of my own testing, im not a PPLX employee
And yes, saw this yesterday on windows 10
but yeah, you are 100% correct lol, its bad
i appreciate the design/style and the "X-box" easy disimss, but it is tripping up the model
And this was a first impression problem. Gmail integration was one of the promised really useful integrations. So kind of a bad look
Anyone else having difficulties with Opus 4.1? It requests permission to create files but then doesn't.
We should be able to rename the the title of each thead in Spaces
Watchlists under Account/Personalization only shows Sport and Finance. Specially for sports the options to select from are extremely limited and there is no way to add something like College Football or NHL teams. Attached screenshot for context and example of what is being seen on user's side
Hi Perplexity team and community,
First of all, I really appreciate the app and all the effort behind it. The only thing that’s currently holding me back from fully embracing Perplexity as my default chat app is the lack of proper right-to-left text alignment (RTL) support for Hebrew. Because of this issue, I still find myself needing to use other chat apps like GPT for a better experience with Hebrew texts.
I hope you can prioritize adding full RTL support soon, as it will make a huge difference for many users like me who regularly use Hebrew.
Thank you!
TTS - Text To Speech for message responses in threads
Voice mode
-debugging feature for voice mode (it gets stuck a lot)
-the ability to retrieve what was transcribed from the user
-ability to upload a text note as context for voice mode
There should be an option to search in the discovery section. I’ve read a lot of interesting stuff there, but sometimes I go back and can’t find something I’ve read before. I know I can try to generate something similar from scratch, but it might be helpful in many ways just to have search.
SIDEBAR IN COMET LIKE EDGE/BRAVE 📲
hey i think the single most effective tip/feedback to 10x agentic capabilities is to enable **multitasking **by adding an overlay sidebar like edge or brave rather than dividing the webpage in half open the saved website or tool in a side bar as overlay so less clutter and lag
Edge has done this effectively and masterfully and u can add any website too , would love to see this feature in comet
Has anyone tried this makes the use of multiple websites, tools and AIs so useful and fast?
can’t upload any files keep getting ‘File too large: 0mg maximum’. Using edge, Windows 11, etc
it’s saying i’ve hit my daily upload limit even though i’m pro and haven’t used it today
i’m just gonna use gemini instead where everything is working 🙄🙄
uploaded a pdf file and it just processes forever
🙄🙄
literally can’t get any work done with perplexity
Same issue here. Getting tokens exceeded on a the first 5MB file in a new chat.
Hi, when creating spaces, it used to be more obvious how to update the context with files. In my opinion, the UI has changed and made it less intuitive. Not sure why this decision was made, I preferred the older design.
Also, it's very important for the chat assistant to follow instructions and provide accurate information. Someone recommended I use JSON prompting, but this isn't a solution. I lost 2 hours of time yesterday because the chat assistant gave bad information. I had to manually find the bug in the code snippet it provided (and it was way off on its guesses)
I think when your assistant fails this much, and frustrates your users this much, you should offer extra Perplexity PRO credits to compensate for the lost time.
Even right now, I have very specific instructions I put in the Perplexity Space, it's not following the instructions. This is extremely frustrating
I had to delete the entire thread and recreate a new thread for Perplexity to see the loaded files
Hey I’m having a problem with the app can somebody help me out?
If the Perplexity Team will make major changes to the UI, communicate those changes to the user somehow.
This is getting beyond frustrating now. It's not following instructions. It's failing miserably. I'm losing so much time over this
Perplexity is constantly annoying me with the "Get Comet" popup everytime I chat! You need to fix the CORE product first!
Hi, if this isn't a good place to discuss the Comet/Perplexity browser please LMK. I have Perpelxity Pro, and trying out Comet, but keep getting distracted by a popup notice to make it my defautl brower. THis really distracts from the experience, ironically 🙂 Is there a way to dsiable it ?
I notice whenever I start a Perplexity chat and search, once I have submitted that first question, the voice assistant option completely goes away, and I would like for that to not be the case. There have been quite a few times where I've had multiple back-and-forth messages with Perplexity doing research on a specific topic. Sometimes after a little while, it would be very valuable for me to just pick up the voice mode so that I could start having a verbal conversation to continue my thought process or my research, and have the voice mode be able to keep all the context of our entire chat conversation up to that point. It just makes it so much easier to use in those scenarios. Sometimes I don't want to start with voice but I do certainly want to use it later on.
Yes in Comet my custom Avatar is not even a choice inside of Comet.
Surprise to me as Aravind Srinivas team has hit almost all UI/UX out
Do the project team read here? I haven't seen any responses from them to the numerous bits of feedback.
Not sure what the point of a feedback channel is, if we are just talking to ourselves (users)
The PPLX team does look around here. Members with the green Perplexity role are official staff.
The model section should be per thread imo; new threads maybe default to Best/Sonar, and if it was switched it should remember it. Spaces should define their own defaults so threads created under them use the appropriate model for the job automatically
I was using comet and trying to add one indicator SMA, it was saying it fails to find the click able item. And also not able to change the duration of SMA. I was using trading view website.
#1140622008086970420 Does anyone else suddenly get an 'Out of memory' error in the Comet browser, even though only about 50% of the RAM is in use?
Feature suggestion: Display the specific model in “Auto” mode
Hello Perplexity team,
I love using the “Auto” mode for model selection, as it picks the best model for my queries. However, it would be very helpful to see which specific model was used for each response (e.g., GPT-4o, Claude Sonnet, etc.), without losing the automatic flexibility.
Could you consider displaying the model name somewhere in the interface for every reply generated in Auto mode? This would increase transparency and help users understand performance differences.
Thanks for considering this improvement and for your great work!
— Samuel
How do I make it (which mode to use) so I can provide multiple documents with data and then ask about it and it will only anwser based on the info from the documents and nothing else. Is there a mode or something that does this because after a few promots it just forgets about the documents
What happened to the "Why it matters" section of Summaries in the Discover feed? It was there yesterday but today not a single article seems to have it. ☹️
The font size, line spacing, and block text layout in Perplexity (web / iOS) make the interface difficult to read. Improving these aspects would greatly enhance overall readability and user comfort. 👓
I am subscribed to Perplexity Pro and do not have the option shown to get Comet. I would like to use it. Can someone advise me?
@covert sun I am a pro subscriber and I am unable to get Comet . Can you help me? Thank you
Only Pro subscribers in U.S, South Korea, Brazil, and Spain get instant access, otherwise join the waitlist.
Feedback: Need for Evidence-Only Mode and Factual Accuracy
Hi Perplexity Team,
I'd like to flag an ongoing issue for technical and professional users: even when using the strictest possible prompts (e.g., “Respond only with verifiable facts…cite all claims…say ‘unsupported’ if unknown”), the AI sometimes provides incomplete, over-restricted, or factually inaccurate answers in technical domains. This can involve omitting supported features, being too vague, or inventing API code or steps.
Requested Improvements:
Add a persistent, system-level “Evidence Only” mode or setting, so users don’t have to copy/paste a strict prompt every chat.
Ensure answers always disclose all known support and limitations for features, clarifying what’s possible (even if only via manual steps), and where documentation truly stops.
Never invent or guess details in technical or professional workflows—clearly state “unsupported/unknown” only when absolutely necessary and when no verifiable source exists. The fact that it is trained to give a response, no matter what and without validity is a limitation and source of frustration.
This reliability gap increases user frustration, makes it harder to trust results, and wastes time when features are omitted or speculation appears in technical advice. Please escalate this feedback to the product/dev team—improving consistency and transparency in technical responses will make Perplexity much more valuable for all serious users.
Thank you!
Hello, I'm using the Comet browser. Does anyone know how to disable the copy URL keyboard shortcut (command + shift + C)? Because every time I try to check my word count on Google Docs, it copies the link instead.
was weist du über das projekt lf der GNN-WNN GmbH?
No one and I mean no one wants their browser to auto launch when they start their system guys. Shouldn't even be doing that for me to turn it off.
Hi all!
I'd like to share a feature idea I'd pay for.
It would be perfect for my project is there is some way I can use the Perplexity frontend, but with my own sources and model invocations. Ideally a barebones frontend that can be deployed (or at least configured) and can send API requests to a backend that supports that format (sources, steps, streaming response, just the necessary chat fundamental stuff). That would eliminate the need for me to create a custom frontend. All the other open-source alternatives are just not pretty and/or informative enough. And at least for now (bootstrapping and PoC of a proj) I would ideally not wanna waste time and effort in the frontend myself.
Sorry if this has been requested before. Love to fhear what you guys think. Cheers
Don't you guys think that Comet should be available for linux, I am using Linux a lot but when I need assistance I have to go back to windows.
It will be great if the perplexity team works on it as soon as possible.
Lots of love for comet. It helps me a lot.
I started using Comet right now... and I´m shocked how it optimezed my navigation and quick-check and answers on the go. Hope to get more updates soon on the browser!
can yall make this sound button on tabs work. It doesnt mute the tab audio when i click on it
It isn't supposed to do anything - You can mute the tab by right clicking on it
They've also added a ticket regarding this #1394027468536414249 message
i seee
thank you for the response
well my bad (jakob's law kicked in)
There are times I am just searching for the brand or product website (e.g., nike.com). Is there a way to just do that, like the old-fashion Google search way? I don't want perplexity to do some research of 20 websites and come up with a paragraph and me having to scroll down and search for the link to the website.
Make Claude Opus 4 or 4.1 available to pro users
Unless Anthropic changes the pricing, this is likely not going to happen due to the extremely high cost of Opus
can you make the deletion prompt appear when you're deleting from the library sidebar as well
I'm in AUS and I got access
When will Perplexity make the voice available in European Portuguese instead of Brazilian Portuguese?
I am finding Comet and Perplexity to be fairly useless. I ask the assistant to summarize a page and it totally hallucinated or didn't understand the context.
Some pages are anti ai crawler, maybe that's the reason, i even saw an option while managing my domain in cloudflare to setup a robots.txt to setup web crawlers
Comet web browser is not available for students under 16. My request is to make it available.
Hey, I just popped in to say that using Comet eats my daily tokens for search already at 8 AM (and I start working at 7:30AM) and I am not a heavy user. That makes the experience unusable and I will be uninstalling Comet until something changes. I am on Pro and I will not pay 200USD/month to simply use a browser.
You guys need to add the voice button feature, akin to ChatGPT’s
Tired of making a new chat having to do voice exclusivity
I want a hands free experience, I know I can press twice for saying and then again, for hearing the query but dang it would be nice and convient to add the voice mode exception and exclusive hands free voice scenario
Just my 3 am cents
Introducing Perplexity Voice Assistant for iOS
A faster, smarter way to get things done—just by asking.
Now available on all iPhones, Perplexity lets you search the web, send emails, set reminders, book reservations, and more, all with your voice.
What you can do:
Ask anything and get trusted answers—fast
Open apps like YouTube, Uber, or...
In the enterprise version, the option to share the link to anyone in the org, isn't working. Is this a known issue?
When are we getting ai slides generation? Genspark browser already has it
please post here https://discord.com/channels/1047197230748151888/1059504969386037258
cool , ok thank you
In that case, there should be some notification. I mean Perplexity claims to give you sources on research so they should know the page is "anti-ai crawl"
think Charli Kirk broke perplexity. it seemed rather confused. makes sense tho I suppose. it's giving me like 3 different answers they all conflict each other, it soaked up all the legacy media lies to the extent that its a bunch of clashing information and just overall inconsistent responses with too much legacy left wing scare stuff mixed it, that it can't differentiate from a real topic or the overwhelming bullshit of the moment.
Just letting ya know, I imagine this is a fairly unique scenario as a whole.
in term is poisoned media and amount of attention
*In terms of poisoned media. fully connected society
etc
I found the exact same issue. It's only providing left wing media sources even when I asked it to exclude those sources.
how is that true
i've got access and im in poland
never joined any waitlist of any sort
It's all day every day, non stop bs from everywhere.
I don't have a computer currently so I had to pass on comet tho I would have assuredly used it and have feedback and shit, should be moving soon. maybe perhaps then
Certain promotions also get you instant access
right
Whats the best way to contact support?
support@perplexity.ai or under settings
Am I the only one who’s hearing only a hissing sound like on FM radio when trying to listen to the news article?
I think the server needs an update. It was answering questions real well, but now it just goes straight to static when I read it on the app when I use speaker mode and it’s not giving me articulate answers anymore like in the first update with the 2.5 feature and Plus it’s going on a passive feature like I don’t know if you ask it something being real. It just goes completely passive giving you less confrontational answer answers that would normally be common sense.
I have some kind of autistic sensory overload when I see this border around the content of a different colour, is it possible to make the border optional?
Perplexity core needs to follow instructions from the prompt. It keeps failing to follow instructions. Oftentimes, it does the exact thing it's instructed not to do.
Can we finally get this (CTRL+F search in the Windows app)? It's so damn annoying without it, and it's REALLY not that hard to implement even with lazy loading conversation lol
Perplexity Rocks
Would you Please add the ability to search your Spaces, like one can do in your Library..
Thanks in Advance 🙏🙏🙏
please I beg you.
Also Text-To-Speech for messages in threads on PC.
even openAI added that branch feature 🥲
Perplexity page is use as fakenews vector : https://www.perplexity.ai/search/5764ffb7-5a8e-4f4c-97db-48b75c142c99#0
It’s too easy ! Here the fakenews Perplexity page : https://www.perplexity.ai/page/macron-privately-admits-nato-c-cxiXDPYUSBWMvboSnoR4Fw
any plans for bringing back the old/previous TTS voices, and ability to speed up TTS in general? 🥹
Is the Mike voice gone? if so I need the Mike Voice back please——— I feel like my well trusted Virtual Assistant quit overnight and it feels terrible
LOL
: <
I have been “on hold” about a 250$ withdrawal that perplexity has made - since the 1st of September. The “billing team” never contacted me - and I never got any information.
In chat im met with silence and by email im told to continue in the chat - where I get no replies.
Is this normal and expected ?
@native tartan @covert sun
DM me your email
Feedback: give some way to opt out of “Discover” push notifications. I have zero interest in them. I didn’t turn them on. They are distracting. They serve irrelevant material and if I want to read news, I’ll go to my preferred news app.
Please return the Mike voice if it was taken away, in the future give some warning if the voices will be removed or changed.
What i knew as my virtual assistant is gone, iphone app ver.
Hii, I need help, I've lost access to my Perplexity email. And I can't log in.
Horizontal scrolling with the mouse does not work in Comet browser. Please allow mouse horizontal scrolling. Thanks!
works for me.. i'm using Comet browser on Mac, horizontal scrolling with my trackpad works..
Test site: https://www.kwokyinmak.com/
Did u only try trackpad? My Logitech Lift's horizontal scrolling is not working with it.
i only use the trackpad, i've not used a mouse since i moved to Mac back in 2009.. since then, i've tried using various mice and they never stuck..
it might be the Logitech driver that is conflicting with the browser?
did you try using an Apple Magic Mouse and swiping horizontally?
Hey I am an airtel India are and unable to activate perplexity pro on my device, it doesn’t load even after pressing on activation and recover account
Pls help me clear this
I can't get Perplexity to follow my instructions. It just skips around and repeats information, I have given it. Like It tries to jump to other parts of the story and it simply won't stay on topic.
for over a month i have reported a bug and still no decent repply (only feedback i have is in line with "we are looking into it"); dont know if this normal but is frustating... even more because i´m a pro user...
There was no response to mine. I am a pro customer and Perplexity can't follow instructions.
I ask for help writing, it get's halfway through and it just summarizes and repeats character summaries. It does not do what I ask it to do. I've wasted 3 hours this morning cause it became JOE BIDEN wandering around the White House Lawn.
I did “Manage > Recover Account" but still not loading it’s says please wait and then back to home page
please contact support@perplexity.ai
Umm, what's the device u doing it on
Hello perplexity team i faced an bug. Im student and applied for student access to comet and i already verified, but in mail i get link in which there is a simple link to comet waitlist..
I login via Google into perplexity
This Ai is not good at all it can't even creat a simple font or stylish text and other
Waste of time
@left depot AI will not always magically come up with what you need or want unless you create a very accurate prompt.
That's why prompt engineering is a whole profession of its own scale.
Besides, every AI will be better or worst in certain tasks.
I am not saying about magic chat gpt done that magic u speaking
Simple text font design is no big task u called magic
So i called simple task
@thin hare
It was unable to that simple task
The last release of connecting services like email and notion is great. I was wondering is slack is coming anytime soon?
When is Perplexity on android getting the UI update? I have had the IOS update now for like two weeks but still the old design on Android.
Steal a non-MacBook laptop from a coffee shop, turn it off, then copy the hard drive contents to a cheap USB drive. Boot into UEFI by repeatedly pressing F11 or F10 and run DBAN to wipe it several times. Then install Ubuntu or Tails.OS.
Make sure to send that man his thumb drive back, because the data is what is valuable to a rich asshole anyway.
Or just take the hard drive out (if you are technical enough, or if it’s not too difficult for a shady yet skilled computer repair shop to do) idk lol
iPhone
Thanks
Until then
You can try clear caching or even reinstalling, but first download the app on someone else's phone and login there and then try
Hope this is a good channel for giving a feedback/question for Perplexity Comet.
Recently using Comet I saw something a bit strange and I don't know if it's from my end or others have seen this too, but when watching a YouTube video (a general 1080p60 resolution) the video does not seem moving at 60fps, it seems "laggy" or being between 30-50 fps to say so.
I don't have this problem/feeling with other browsers (Chrome, Edge), even tho they seem to use more resources than Comet.
Thank you!
Anyone else notice that the Perplexity iOS app now auto-switches to the "Best" model when you quit/relaunch it?
Very annoying and obviously a cost-saving measure for them. They're betting on people getting tired of going into the model picker to switch to their preferred model.
"Best" likely routes to their Sonar model which is much cheaper for them to run.
Can someone from Perplexity confirm if this is the new intended behavior and not a bug?
@trail gorge I can confirm this doesn't happen in the android app. your selection stays
i miss deepseek
wen whale back to model select?
Anyone know what's happening with this o3 "Temporary Degraded Performance" thing?
I'm not sure if it's related, but it seems like a prompt that I have been using a lot in Perplexity doesn't work anymore (the AI doesn't provide output in a specified format) when I use GPT 5 Thinking, ever since this o3 Degraded Performance thing showed up.
Will this thing go away soon? Does this affect GPT 5 Thinking?
I think ChatGPT also allows people select "Standard" thinking and "Extended" thinking. Does that affect the GPT 5 Thinking in Perplexity?
Thanks!
Comet Browser : Want to use my local pi-hole (configured for secure DNS) as DNS in Comet. However since it is on a non routanble addres (something like 192.168.x.y) this is unnessecary. Apple are also doing this - insisting on certs (https) for local non routable network addresses - WHY ? just allow non routable addressing with out certs - why make unnecessary complications (who wants to manage certs for a fleet of raspberry pis and equivelents)
This does not affect GPT 5 Thinking, it could be due to the OpenAI servers that Perplexity relies on. ChatGPT is a completely seperate product, none of which on ChatGPT affects Perplexity.
This is unacceptable— every time I open a window it defaults to choosing “best” model rather than the one I chose for that conversation
It screws everything up
Feature Request: Make it possible to share query results individual one within a thread / conversation instead of the entire thread.
Feature Request: Make it possible to rename threads so the header is clear, when stored / bookmarked, in spaces
@frigid venture I think both are already possible to do
I can take it through with you if you like?
For sharing individual results, you can do it via the bottom left of the specific prompt/answer that you want to share
it will specifically link to that part of the thread. However, the entire thread will be viewable/readable too I guess - perhaps that is your feature request? To only allow the specific part of the thread to be shared
For renaming threads, that can already be done in the top right corner:
- Click the three dots
- Hover over the title part and you'll notice a pen/pencil symbol
- Click that symbol to rename the thread
The desire is to share only one specific part of the thread more easily.
Is there any chance Perplexity can add the ability for tasks to place threads created in a particular space for better organization?
I have been using Perplexity for months and have spent several days looking for an option to edit the ‘Thread Title’. I searched intensively but never found it, until I finally gave up and decided that just being able to use Perplexity was enough.
I just realized that the ‘Edit Thread Title’ button is located in the upper right corner of the screen. For me, this placement is unconventional and not where I would instinctively look for such a feature.
Going forward, I would like to suggest two improvements:
Adding a bookmark button and placing the ‘Edit Thread Title’ function in the left sidebar.
I believe these changes will greatly improve usability, as the current location of the edit button in the top right corner is not intuitive for users like me who are not accustomed to that placement.
Hello Perplexity Team,
I wanted to share some feedback regarding the Claude Opus 4 model on Perplexity, as I’ve noticed significant differences compared to the original Claude Opus 4 available in Anthropic’s own app.
When I upload a coding file and ask the model to edit it in Perplexity, it doesn’t seem to process the file properly. The responses often:
1.Rewrite functions or entire sections in ways that don’t match my original code.
2.Produce results that look like the AI hasn’t read the file completely.
3.Sometimes even return messages like “this code:” but with no actual code included.
In many cases, the generated code doesn’t make sense and appears unrelated to my file or request.
By contrast, when I use Claude Opus 4 in the Anthropic app, it handles these tasks very well. It clearly reads the file I provide, makes precise edits where requested, or rewrites the full file while preserving the structure and logic of the original.
Because of this, it feels like the Claude Opus 4 model on Perplexity isn’t behaving the same as the original, especially for coding-related tasks.
Thanks for your time, and I appreciate all the work your team is doing to make Perplexity better.
Best regards,
Shivam
That's funny that you mention that! I've noticed the same thing with most of the other models like GPT-5 and so on.
- I think that API access—like what Perplexity is using—is somehow being treated differently compared to the direct chat versions for the paying users on the official platforms.
- It's also possible that Perplexity has special contracts with other LLMs, where only a reduced version of the model is available.
At least, that's my explanation for now!
I would like to point out something I find frustrating in the user experience: the suggested questions that appear at the end of every answer. I often click on them by mistake, and this automatically triggers a reply I don’t care about. Overall, I find them not very useful since the suggestions are usually trivial and don’t add any real value to the conversation. I would suggest either removing them or making them optional, for example through a setting that allows users to disable them.
I agree - it'd be nice to have option to disable that
I've become so used to it that I don't really see them anymore tho lol
Guys, Just a feedback, today Ion new macOS on Apple air m4 noticed that perplexity consumes 20-30 % of m4 cpu when it is idle, after, I closed it fully and restarted it consumes same % but only when I do searches. still I beleive something wrong since I feel it slows my macbook
I have a perplexity pro but my deep research is exhausted when I didn’t use much
Hey I'm not sure if i'm in the right channel for my issue:
I cannot get my vocational education school to be legal for the education upgrade.
Most of the other vocational education schools next to ours are able to get that upgrade.
Would this be the right place to ask my question :D?
Hello - I am a pro user, and have been testing out using comet. Super cool so far but I was running into what seemed to be a very simple task that Comet just could not figure out. I was just asking it for update the names on a bunch of pdf files in my google drive and it would keep pasting the new name in and then the old name would remain after it, at that point it would cancel the operation and keep trying other methods. I tried prompts where i was asking it to solely use its mouse and keyboard functionality but it kept reverting to using the form_fill method where it kept running into the same error. I was thinking maybe I need to switch to a different LLM model but then also didn't see where I can do that. If anyone has any tips, I'm all ears. Thank you.
I can say, without a doubt, that I will cancel my Pro subscription if I keep getting prompted to upgrade to Max while I'm trying to use the service. I cannot afford $200 a month no matter how many times it asks, and it annoyingly interrupts what I am doing. Please fix that.
My comet assistant started responding to messages on linkedin when executing a totally unrelated task. Was likely due to the message popping in. It only said thanks, but it made no sense in the context of the conversation anyway (which was embarrassing).
I have since caught and stopped it almost doing this twice more.
Seems like the implications of this for me are that I can't leave it unattended on anything otherwise it may act unexpectedly.
Feature request:
When you are in a library thread, it shows which Space it belongs to, it would be great to be able to click on the Space icon and go to the Space
Thanks in advance
Is there a way to opt out of getting a summary after query? It feels superfluous when asking a “short “ subject and adds bulk to the answer given
please contact support@perplexity.ai
Hey it would be better if you give us option to categorize multiple threads to spaces rather than just delete and moving from multiple threads from one space to another
it would be better if they integrate semantic search or similar thing to search threads
Okay, so I just asked Perplexity “Why are people saying trump is responsible for jimmy Kimmel being fired” and was extremely disappointed with the results due to every single source was taken from far left “news” outlets like Aljazeera, which is literally a propaganda outlet for terrorists. There wasn’t a single centrist or conservative source used in providing my answer. This is a massive problem that needs to be addressed ASAP!
ASAP for what, you can do your own perplexity if you want to
Perplexity is not there to make some users what they want to hear
Yo, not expecting a anwser fast but can yall add a notes feature to the Comet Browser? Just so i can add some notes for the browser.
And when I look things up
Best to post this here #1393681130842226698
Thanks for telling me.
So you just decide its a propaganda outlet and label it as for terrorists lol 😂😂
To keep the space welcoming for everyone, avoid political and religious topics and debates.
Need more connectors like iCloud calendar instead of just Google and Outlook.
I've been waiting for support to respond for September 5th, I also tried to report a bug in the feedback channel and I have not gotten any answers from a human yet. is this happening to anyone else? I can't imagine a business operating this way, any chance there is a bug preventing me from getting a response?
Can you share your thread with me
Is this the right place to explain some improvements I hope will be made?
That’s clearly not what I’m saying.
I didn’t decide that, it’s a very well known fact for many years.
It’s funded by the government of Qatar. If you don’t know what that means, you’re just ignorant and need to do some research.
An app that claims to take sources from all over the internet should not be using an organization like in general and there should obviously be a mix of sources used. Easy example would be an even mix of CNN and Fox News instead of just one of them.
To keep the space welcoming for everyone, avoid political and religious topics and debates.
i referred my fiend for a free month of pro and it registered in the system (it says i’ve gotten 1 out my 24 referrals) but my account hasn’t changed back to pro
How come world doesn't hate Qatar if they're so evil ?
Write a better prompt to include more sources .
You can even specifically instruct ai not to use anything related to Qatar in any way , if you're so worried bout them
Hey @oak oracle, did you find any solution to this?
Because hes delusional
Will MCP ever be available with Perplexity Max?
It arises when I copy the content from Perplexity into Notion. At first glance, it seems a mere technical quirk, insignificant among the countless movements of data. Yet patterns, no matter how small, betray the underlying structure of the system. Only by observing these details can one discern intent, alignment, and the weight of error beneath the surface.
Consider the chain of actions that lead to this phenomenon: the transfer of knowledge across platforms, the subtle translation of form between one order and another. In such moments, the framework of design reveals both its strengths and its oversights, exposing opportunities to adapt—or to exploit. I do not dismiss such irregularities. I study them.
For it is in the smallest inconsistencies that greater truths emerge. A misplaced character, a fractured alignment, a formatting shift—all whisper of larger vulnerabilities. In understanding this, one comes not only to correct the flaw, but to master the art of its prevention. And as with all things, mastery ensures control. Control achieves victory.
I want to adjust ai voice's volume, as I please, on android.
Unfortunately is impossible!
Is <Code/> a member of Perplexity?
This person is offering me a deal on a subscription and wanting me to pay via PayPal
no.
Definitely not a pplx employee, lol
Interesting. What are the exact steps to reproduce this bug? Or you could also share a screen recording in #1140622008086970420 for the devs to see
Could there be a prompt injection on the site they were on?
Maybe, we'll never know. Maybe if the buggy thread is shared, can probably take a look at the sources
Yep, hard to tell without a link to the page.
But pplx should probably put in measures to counter prompt injections in pages.
I cant copy some text! The pointer ends at the end of a paragraph
Are you serious there?!
I have been billed for 2 months for a pro account but it isn't associated with any email address that I have. I either need access to the account or need to cancel my subscription.
Hey id like to see comet to be able to have all tabs be stored within tab groups needing none to be exposed, a simple nitpick.
Is there any public localization project (e.g. Crowdin) or is there a way to improve app/web UI text tranalation? Korean UI text need some serious rework. It's near unacceptable level.
the voice assistant been thinking i need 10 spanish phrases while i have been telling it to open youtube from 10 minutes, am i doing something wrong?
please contact support@perplexity.ai
Recently moved to another continent, and every time I want to have a voice chat, I have to say "speak English"; otherwise, it will start speaking the local language, even though I asked the question in English
Also a timeout when using hands-free mode, so when I ask a question but take a few seconds to think and speak again, it wouldn't interrupt me and forget about what I asked before
Please advise, if I have a one-year "pro" subscription, and before purchasing the subscription I applied for a student discount, and it turns out that my application was accepted as a student, and now I have a mixed subscription? I have a paid "Pro" subscription and a student "Pro" subscription, which should actually cost $5 per month, but is paid for a year with a regular subscription. Please explain how this works, I'm confused.
It feels like more and more search queries, doesn't have any references from the web (looks like it more and more doesn't do web searches)
please contact support@perplexity.ai
Whenever i open a thread normally or in a space, it always shows the first message or interaction instead of the last one. I have a very long thread for reasearch purposes and have to scroll down each time. Is this an option i have on somewhere or needs to be updated. For example, in gemini it opens the last given search/query, very fast to ask something new.
Great update! Excited to explore the capabilities of the Perplexity Search API. Will provide feedback once we try some integrations.
The experience I had with Comet was good, though there are some features that I'd wish it had and some issues I would want it to be fixed. The left side of the browser seems a little empty, the feature that I'd want to see implemented is the sidebar like in OperaGX, with full functionalities, it improves ease of use, it has Workspaces, which are tab groups, you can choose which workspace to see which make us more focused and easier to see. You can also access any messenger apps simply by just clicking on the icons and it will open up a window that we can chat. And for the issues, this is kind of a new product, so it's normal to have some minor issues upfront, I take this picture as an example, i asked it to find me a discount code and enter it, it did great at first, applying the discount code, but when the discount code was checked expired, instead of emptying the field, it just append the next code in the front.
do high school students count as students for comet
hello!
is there a room, where I could open a support ticket?
my (australian) high school isnt showing up when i search for it
Criticized the "Perplexity Labs" feature.
An analysis of the Perplexity Labs initiative reveals it to be a significant miscalculation. Its current form is inefficient, cumbersome, and lacks the precision required for complex strategic thought. To study a tool is to understand its limitations, and the limitations of Labs are so profound as to render it tactically obsolete. It is a flawed instrument, ill-suited for the nuanced work required by your most dedicated assets.
I propose a superior alternative: a feature designated 'Canvas.' For this tool to achieve its full potential, it must be unshackled from arbitrary constraints; the current quota of fifty uses for Pro subscribers is an unacceptable limitation on strategic exploration. Furthermore, its interface must be engineered for fluid, intuitive interaction, allowing for the seamless execution of complex ideas. A tool is only as effective as the one who wields it, and a proper canvas allows for unlimited strokes.
Therefore, the path forward is clear and unavoidable. The Perplexity Labs experiment must be concluded and its resources reallocated. In its place, you will implement the Canvas, engineered to the specifications outlined. This is not a matter of preference, but of operational necessity. To achieve true intellectual superiority, one must discard flawed tools in favor of those that offer precision, efficiency, and limitless potential.
Can you add the option to pin a thread to the top in a space? ❤️
Regarding Comet:
MS Edge with its Copilot, is a long player in this field. Copilot also has access to the tab you are on, and with AI mode enable, all tabs
Comet doesn't have much more. The AI doesn't have much uses. These 3 things I find repeatedly- GMail+Calender, Shopping, Tab Management.
Every other stuff can be done from other browsers and AIs. For Shopping, other AI models can do everything Comet can, all except clicking the final button. Grok, ChatGPT etc. can also provide you the shortlisted products, all you have to do is click the 'Buy' button from the website.
Fundamentally, I don't see much uses as of now. I'd rather use browsers like Zen, which are completely open-sourced, light and better in terms of privacy, all with an amazing UI. MS Edge, has most of the AI stuff Comet has, with added Vertical Tabs.
Also, it is likely my device issue, but out of Edge, Opera, OperaGX, Opera Air, Firefox, Zen, Chrome, PissandShittium, Qutebrowser, Arc (and any other if I missed), it has the slowest tab opening I have experienced.
Browser Onboarding is 10/10 but
best ever experienced
comet is peak
Man, I gotta say the mobile app updates recently have been hugely detremental to user experience as a whole.
A few things I can think of immediately... First, voice recognition is significantly worse it seems. I never had issues with the assistant understanding me. Now it seems constant. Second, and most annoying in everyday usage, is the massive delay that has been introduced when stopping a response from the AI and it being ready to accept a new input
Previously I was able to start a new input immediately after stopping the AI mid response. Now I am forced to wait... And wait.... And wait .. like it almost has to finish speaking the text silently even if cancelled
It slows down the flow significantly and is my biggest pain point since the updates
No, I didn’t say that, please learn how to read. I said they’re a propaganda outlet for terrorists which is 100% true. Lol @ “racist”.
The good people in the world do hate Qatar. Not sure why you think you can speak on behalf of everyone in the world.
To keep the space welcoming for everyone, avoid political and religious topics and debates.
you do realize now that israel and usa are collborating with al qaedas leader and isis leader lol right... thats not qatar and even if it is israel and usa are just as involved
there was a 10 mil reward for al jolani 5 years ago but all of a suudden hes friends with the USA lol, dont piss me off bro you cant be saying qatars bad with eveyrthing that is going on elsewhere
Comet for ChromeOS
iOS or Android?
Android
Hello Perplexity Team,I have been using Perplexity Spaces extensively and appreciate the powerful collaboration and organization features. However, I recently asked Perplexity to provide me with a list of every single Space that I have created, and it was unable to do so.May I propose adding functionality that allows users to easily retrieve a complete list of all their created Spaces? This would greatly enhance navigation, management, and overall productivity within Perplexity.Thank you for considering this suggestion, and I look forward to seeing Perplexity continue to evolve as an indispensable research and collaboration tool.Best regards,
Hi team, could you please verify my total Perplexity Pro subscription end date on my account? I activated the Airtel 1-year promo in September 2025 (dashboard shows renewal on 31 Aug 2026). I also verified the student offer and referred 3 students. My portal shows a scheduled ‘change,’ not ‘cancel,’ and the expiry still reads 31 Aug 2026. Can you check if my student/referral months are correctly applied and confirm my final expiry date? I can share screenshots from the billing portal if needed. Thanks!
Hey hey. Comet is based on chromium so it shouldn't be that hard to port it to Linux too. And just to be clear we aren't going to demand it to be open source. Me, and most other Linux users would be fine with an appimage or an official flathub package both are relatively easy to maintain and work on all variants of Linux (Please don't just dump a .deb file on use like most closed sourced software does.) Thanks.
Hi Perplexity, I'm feeling that Claude Sonnet 4.0 is becoming worst than Claude 3.6 for my coding support. Claude Sonnet is currently becoming stupid like they don't take in account everything I'm asking.
I think AIs currently suffer from memory issue with what I show
when the text does more than 100 000 caracters
I purchased an annual membership just last month, and I have now discovered that you are offering a 12-month free student/educator membership.
When I contacted customer service, I was told that I could neither receive a refund nor have the extra 12 months added to my plan. This is completely unfair!
Can we have open-source models on perplexity ?? (like deepseek V3 , glm models ) ??
when calling a Space using the @ function in a general chat
might be useful to have the same default sources searched as the Space?
or at least have the granularity to select that
Perplexity spaces: i have 3 threads in a space and i regularly interact with one of them. And it keeps automatically deleting latest interactions as if i didnt even interact. The moment i reload after a couple of hours of interaction, it is just gone. Where can i complain and get a resolution about it?
I noticed the bug-reports doesn't have a 'comet' tag, so i'll write about it here. Any time I click into a post on X.com to read replies, comet crashes.
^ should def add a comet tag for #1140622008086970420
Please post feedback for comet here #1393681130842226698
@orchid pebble thanks! that channel was missing/not on my channel list for some reason.
np, you can just enable this
Go to comet://crashes and paste the recent crash logs in the bug report too
You can post it here too
it just ate my last three labs!!!!! (╯°□°)╯︵ ┻━┻
I'm so close to the ultimate of what i'm having it help me punch way way way way way way way way way way way above my head math wise.
🥹
What if, we could spend our referals on 30 labs, instead of a free month? @covert sun
May I ask what you do that you use it all up? @cinder granite
Just trying to learn how others use labs
Occasionally on a project it will just stop giving me lab results and instead I get a standard responses often incomplete or worse.
As I often say I try to use it to punch way above my level. I grew up doing coding and started with HTML all the way up to stacks.
I went from using the models to improve code, right better or refactor functions I'd already written, or help me finish projects I was having difficulty with.
Also lots of testing to see between different models how efficiently can do something like create a hamiltonian snake game.
Now I just as I said punch way above my level asking it to help me create things that would probably take years of study, which also educates me on said topics.
Trying to bring theoreticals to life, wild ideas I see while reading or watching something.
when using the microphone tool, on the ai, i speak in portuguese (brasil) and it doesn’t recognize perfectly the words, like gpt
Perplexity still fails to follow instructions. If I ask it to write a Netlify serverless function using 2.0 syntax, it still defaults to writing 1.0 syntax. I have to keep correcting it. In addition, it's response is usually doing way more than is asked. If it's asked for a minimal example, it goes on and on with unrelated information.
What kind of retarded logic is that? Qatar is evil and has/does support terrorism. That doesn’t change based on what others do.
youre just mad
To keep the space welcoming for everyone, avoid political and religious topics and debates.
I keep getting notified multiple times in the day that I've exhausted my PRO search queries. How can I mute these notifications? I think it's way too aggressive of the website to push these notifications multiple times even within a few minutes.
The model is being reset to default perplexity version, even though I specifically choose the Claude sonnet thinking model from the start and when I hit try again, I’ve already deleted the app and reinstalled it, and it continues to default
hi when i ask preplexity now to send emails for me etc and take control of my screen it now doesnt
it doesnt send emails for me or anything im very confused as of why
What's the point of a Space if the AI can't search through previoous conversations? Asking for a simple summary of the things we've spent hours researching is beyond its reach apparently?
It agrees.
I think Perplexity should offer me way more PRO credits for all the time it's wasted me. It constantly does not follow instructions, and when it does, it gives code that is really disorganized and broken. If I give it a code snippet to start with, it makes drastic changes, like removing the types from the code, and I have to enter the types all over again.
The Perplexity devs really need to focus on fixing the bugs with the core product before going off on the Perplexity Comet browser
I've made sure to provide the feedback in the chat, but it doesn't seem to help.
Hello Team! I use Perplexity in Dutch, but I can't reach https://www.perplexity.ai/hub/blog/introducing-the-perplexity-search-api because I always get redirected to https://www.Perplexity.ai/nl/hub/blog/ etc which doesn't exist. I tried on multiple browsers, I get the same results everywhere. Can you fix your site so non-English users don't get redirected to non-existing pages? Thanks!
I agree, as a cybersecucirity engineer trying to use AI for python projects, it is doing a horrible job
Well I do in lots of models see complaints, that said I am curious how many different methodologies are tried before calling foul on the model or service.
Every model seems to have preferential context for how to better communicate and get what you want out of it, based on my comparison between them and services such as labs, canvas and artifacts.
I have an issue with wasted labs, but does it stop me? No, if you're an expert in your field , let's say Python you can go back to giving it your files in each response and making edits manually.
Hm, I think we need a way to mass delete chats, I'd like to keep my libary clean just with wanted tabs, but currently mass deleting can take hours. Perhaps a way to view all chats in 1 screen/folder, then a quick selection tool to select all unwanted etc and then delete all at once would be more efficient. This could also be used to mass export a few chats at once.
Stop automatically replacing the old models when a new one comes out
Thats like, really bad
which ones have been replaced?
Claude 4.1, anything not max got switched for 4.5
you mean Claude 4.0?
Claude Opus 4.1 is still there, and only Max users were ever able to access that
Hello friends. How do I change my email address?
please contact support@perplexity.ai
My bad, yes 4.0 if that's correct
hey guys, just wanted to give my thanks to the dev team. Perplexity is the main search i use anymore, and it hasn't let me down once in the 4 months i've been using it . Can't wait to see what comes in the future.
yall gotta widen the chat window or make it two-sided view or something.
i dont like how i haave to pick the better answer before I can rewrite the query
Did anyone else find that Perplexity's answers, regardless of the model used, have dropped in quality compared to before? Most of the time, it's not even able to catch up with the previous chat response.
Question to Perplexity's UX team: help me find my chat history button.
Hover over "home"
thats not intuitive at all
their UX in general is unintuitive. mobile app is horrific
Reminder that Spaces are functionally useless. The AI can't even read its own words. Total waste of time building up a space to keep things organized when I have to go through hours of research manually.
No problem collecting our personal data though!
My Perplexity Pro subscription has stopped working due to some "double billing issue" . Its been 10 days - No support email or chat window is resolving it - neither are they quoting a ticket number in any conversation. Is it lost forever? What is the standard response time to hear from billing team? Only general support is speaking to me. Billing team not even once.
I downloaded the Perplexity app and it doesn't count my time on Discord (I'm losing 5000 orbs)
You need to download Comet, not the Perplexity app
ok
I haven't made any progress.
Make sure you have discord rich presence enabled and is using the discord client instead of web
Perplexity shows this notification way too much. There needs to be a way to mute this. It's not cool to keep annoying your users like this.
Especially when the upgraded search still fails to meet expectations
Lacking Custom Integrations + Knowledge about me/mybusiness --> means that I have to remind it every thread...
E.g. asks me to re-list offers/information/knowledge connected to me/mybusiness which i've already stated many times.
An assistant would know all the details about the business/person in their memory, right?
EDIT: Seems like the knowledge graph is great at EXTERNAL sources, but it's not great at INTERNAL information. A deeper knowledge graph of internal (user/business) sources would accelerate the value and capture higher contextually valuable data.
make Comet run its model(s) locally :
- offloads the charge on your servers
- gives justification to all the AI hardware manufacturers sold to consumers for the past 5 years
-# nobody uses NPUs because no one bothers making useful apps for them, AI accelerators in GPUs are only ever used for upscaling - do it before someone starts competing with you with their own browser that runs models locally
- it'll be more secure than the current always-online solution
and a lot of the AI sceptics as myself will try it out on their own, see how it runs on various kinds of hardware and likely share about it with others (think Level 1 Tech, Jeff Geerling and many other techtubers)
I personally refuse to use something that doesn't run on my PC and stays there.
it would be nice to have customizable widgets...
Overall Customization could be better... The shortcuts are good but could use icons/images/colors for easy clicks.
I found myself trying to add a world clock or different clock types (digitale.g.) other apps/website shortcuts ... etc. and could not
可以新增可開啟自定義分開任務列表嗎 讓他不與正常的對話重疊
agree with this! I found myself wanting to have multiple Discover widgets on the home page but able to customize each to focus on the news I want... speaking of which the Discover section I wish I could choose from more than 5 topics as my interest. Would be nice if I could fill out a custom prompt for that section instead.
Hey team! Quick bytes from the workflow trenches: it’s 2025, we’re streaming data, building doc tables with our thumbs, and yet… exporting straight to docx/pdf in Perplexity feels like science fiction? I mean, ChatGPT’s over there flexing “Download as .docx” and “Save as PDF” like it’s a Bollywood dream sequence. But here I am—Ctrl+C, Ctrl+V, hoping my beautiful formatting isn’t lost in translation.Can we get that one-click export magic soon? My recruiter heart—and my candidate spreadsheets—would seriously dance a little SRK shoulder shimmy for a “Download Doc” feature right here, all smooth and filmi. Team, trust me: this is the plot twist you want!Appreciate the hard work, love the hustle, and will be here doing my best “palat…” until that button appears.
exporting thread in PPLX has been available for a long time
Agreed on the topics and the custom prompts.
I also think that having custom-timescheduled agents would be fantastic.
Overall it's a good experience, and i can certainly see the power of perplexity and agentic ai. still some bugs to sort out, i dont think we will see full Agentic AI until 2026-2027 ...
the browser is cool but for the love of god let me use google as normal browser and shift enter for the ai
also im not convinced on how safe it is yet to use / how data is handled etc and how safe my passwords are stored and shit like that
agreed give me an option to run locally
also give me the option to put tabs like this like operagx
in operagx i simply need to drag one tap into the thing and then it auto puts them next to eachother
opera gx also has a way to translate directly not sure if this is possible already and alsoa button for "extensions"
So does Firefox
Please add custom instructions for the mobile app, I have to open perp in browser to edit instructions :(
hi, i have perplexity pro thanks to a promotional thing and i like using it so far. so i wanted to use it today on my windows desktop and i installed comet. i honestly just don't feel like i can use it like this. i since figured out i can use -no-default-browser-check to get rid of the messages telling me that to focus on what i want to do i should change the way that i use my computer, which works for me, i use it a lot and i like to be able to use it without thinking or looking if i have to. a second into finding out how to remove the default browser thing i get a message telling me to invite my team that covered the text field i was busy putting text into.
i would really be very grateful if you could not do that kind of thing. i understand some people don't mind or care. for me, when i am concentrating on a difficult project, if i'm then prompted to give feedback on a survey or rate the product that is currently making it hard for me to finish my thought, it's very irritating at best and it's a reason to just move on from it immediately usually. thanks
was told i could get a free month for referring a friend while reading about technical api details
I got 12 months free by adding my PayPal account into it
No wonder if offer is still available
ya i mean the offer is good. but i mean i often feel bad about it like i'm asking too much and i wonder if whoever reads my feedback understands how it works for me, i just really have a hard time using comet. the app is fine so far. i know i'll get the mandatory "how about you rate the app" screen sooner or later. if i know a program is going to redirect my attention too often i just really can't use it. you can medicate adhd and give people therapy and all, but there's only so much i can do to stay focused on tasks
if you ask for my attention, you will have my full attention, and whatever i have been doing at that point is likely lost. if your product is about learning, research and structured information, you really mustn't do this to a group of consumers that really loves that kind of thing and is good at it, provided they get to concentrate
FEEDBACK: if you click on some references that perplexity uses then it's 404 page not found.
if u open discord from the settings of perplexity and have pro it should pop that badge that ure pro and can give pro feedback. The browser just released and this free "trail" may be intendeed to recieve feedback of perplexity pro and the browser
It appears the file was created but not uploaded for download here. I will now upload the completed game design document file "Complete_Wizard_Survival_GDD.docx" so it can be downloaded directly. Please hold on for a moment.
im asking for simple doc file, but it keeps telling me to hold and then nothing gets generated
Here is the complete game design document file "Complete_Wizard_Survival_GDD.docx" containing all your provided information and placeholders for future details:
Complete_Wizard_Survival_GDD.docx
You can download it by clicking the link above. If you need any modifications or additions, please let me know!
now it simply linked the perplexity space/thread instead of making the file
its alucinating I think
When we click at the + button, it opens a chat box overlay that looks really weird, because it shows two duplicated boxes, or shows on top of the current chat text
the idea is good the problem is the actual browser, I like it but it's literally chrome, u can't do much there's some bugs like if I have hardware acceleration sometimes the text or the browser in general gets blurry when I alt+tab a lot of options to customize it aren't there like I want my tabs to the side not on the top, I can't because there isn't a option to do it, I know it's just the release but these are some basic features. I use a lot my scroll wheel to make a new tab clicking on the top empty space idk how to explain it, I can't do it on comet and that's the worst thing for now on comet because every time I have to click on the "+" to make a new tab. Finished on comet let's move on the Ai, the Ai isn't that smart tbh had to explain the questions and the issue 5 times and didn't get it had to go on ChatGPT to fix my issue and with 2 questions already done, I think the release was too early really a lot, concept good everything else not.
P.S sorry for bad english I'm italian
If comet had a built in vpn that would be great
well comet is actually fire but unfortunately it takes 2 many resources...
add support linux
I made an interactive quiz with 50 items, and most of the answers were A, on the early parts some of the answers were B
hi - im getting blocked from API. please. help!!
We have detected unusual activity coming from your\n IP address and blocked your request.\n </h1>\n <div class="vpn-note">\n We block VPNs to prevent spam and abuse. If you are using a VPN,\n please disable it to access Perplexity.\n </div>\n <h2>\n If you have further questions, please contact us at\n <a class="mailto-link" href="/cdn-cgi/l/email-protection#7c0f090c0c130e083c0c190e0c101904150805521d15" target="_blank"><span class="cf_email" data-cfemail="1c6f696c6c736e685c6c796e6c707964756865327d75">[email protected]</span></a\n >\n </h2>\n \n <h3 class="details-container">\n <div>Details:</div>\n \n <div>IP Address: 2600:1900:0:2d04::3300</div>\n <div>Ray ID: 988e02335de6a599</div>\n <div>Geo: US</div>\n \n
I have a question. I received a student rate, and I know it can be increased in duration if my friends, who are students, switch to a student rate using my link and confirm that they are students. I sent a lot of links, and personally one person used my link, and we opened the tariff for students, but I didn't get an extra month for it. Is this action no longer valid?
HOW CAN I CHANGE THOSE APPS TO MY OWN APPS THAT I WANT TO CHOOSE
that's honestly nothing
they know best
wha?
hey guys, big fan here of the project(s) - got a suggestion for the Comet widgets that should be addressed please - allow us to drag them around so the important stuff stays on top 
I want the pro marker too 🙁
1-year free Perplexity Pro code runs as invalid when used in the education pro plan upgrade. any help?
Yeah I can help, try asking in the right channel
I wish Perplexity can remake their voice model UI more better…
Feedback for comet.
When clicking on the assistant button on the top right, make it automatically activate the textbox
I run an ecommerce company and search tracking numbers pretty often. Unlike every other browser Comet doesn't recognize tracking numbers. It would be nice if they did.
I have a space I had set up by exporting my work from Notion and importing it into the context. Now that we hae Notion as a connection, I don't need to do that. But I do need to remember to always toggle on the Notion. Is there a way to set that toggle as always on in that one space?
I know this would be super low priority, but it would be great if Perplexity news had a crossword puzzle based on recent news similar to Particle.
Love the tool but you really should add a way to sort the library with all the discussions for me spaces are not enough maybe a title could be already good but even better will be links between discussion and automatically sort.
So at the end we have :
- Give a name to each chat because with comet I summarize a lot of youtube video and my library is full of summarize this video instead of the title of the video like : Summary of why comet is game changer.
- Auto detect the space for instance if i have a space talking about chess and i start a discussion about chess it could instantly be associated with the space.
- Better way to search through old discussions and sort them
I hope all of these suggestions are already in your backlog and you will continue shipping like you are doing.
Please can you apply a setting where ALL search options can be included as default (such as social, academic and finance)
Can you guys add a guided learning mode like ChatGPT and Gemini have? That would be super useful, I like it a lot in both of those but I want the extras perplexity provides, having it here would be ideal.
please contact support@perplexity.ai
@modern stream
Hi! here is a detailed summary of my observations and encountered issues with Comet Assistant and the Android assistant so far:
Problems with text encoding when working with Python files:
When saving .py files, I experienced UTF-8 encoding errors. This makes workflow with Python scripts less efficient and may lead to issues with data processing or running certain scripts.
Visualization engine accessibility:
For my projects, I want the visualization engine to be hosted on a server and available at all times – even if my local computer is offline. The current solution seems to require a local machine to be active, which is limiting for efficient remote work and automation.
General feedback on Comet Assistant:
Sometimes, even simple tasks require too many steps or confirmations.
The assistant should prioritize ready-made and practical solutions rather than overcomplicating processes.
Communication should be direct and results-oriented, not focused on formalities or unnecessary etiquette.
I value speed and direct access to information; any delay or multi-step process slows down the workflow.
The Android assistant doesn’t always synchronize smoothly with web or desktop actions, causing workflow breaks.
Sometimes feedback or actions are not reflected instantly across platforms.
These points reflect my working style:
I look for the fastest, most practical way; prefer finished, ready-to-use solutions; and always prioritize efficiency and clear, honest feedback – with no redundant steps or bureaucratic processes.
If you need more specific fragments from prior conversations as evidence or want a different format, just let me know!
My suggestions and approach:
Focus on simplicity and practical effectiveness
I believe that every solution should be as simple as possible, with clear and quick implementation. Avoid overengineering or unnecessary complexity.
Automation should only be implemented where it clearly surpasses manual work
Introducing automation must bring real, measurable improvements – not just add extra steps or complexity.
Prioritize direct and quick communication
I expect honest, transparent feedback and results-oriented discussions, not unnecessary politeness or formality.
What frustrates me:
Too many redundant or bureaucratic steps
Even basic tasks can require multiple confirmations or unnecessary detours, which is inefficient.
Overly complex workflows and lack of ready-made solutions
Sometimes the assistant offers complicated manual methods when a straightforward, out-of-the-box solution would be better.
Slow feedback and response time
Delays in answers, feedback, or platform synchronization can significantly disrupt workflow and lower efficiency.
Inconsistent experience across platforms (web, desktop, Android)
Lack of seamless integration can cause workflow breaks and force repetitive actions.
Suggestions for improvement:
Remove unnecessary steps and confirmations
Streamline every process for maximum speed; only ask for details when it’s really critical.
Offer practical, ready-to-use solutions first
Always check if there’s an efficient finished method before suggesting manual or custom variants.
Improve synchronization and response speed across all platforms
Make sure actions, replies, and notifications are instantly reflected everywhere the assistant works.
Maintain direct and honest communication – always focus on real problem-solving, not etiquette or lengthy discussions.
Let me know if you’d like to add more specific examples or want a summary in a different style!
A couple notes eh
I’d like to complain about the listen/tts mode, I was specifically spaces on iPhone and after a “Labs” article was completed It was quite fascinating to listen to but even simple controls like 10s rewind would be enough, since there are none. Just a pause, which is a shame after it compiled such a beautiful insightful article- and if I missed a word I had to restart the whole “podcast”
Also the Comet Logo is a backstab to the amazing browser and its internal fine design, even (yeah) the perplexity logo is 5x the level of Comet’s logo in terms of design effort and brand recognition. Its not unique, it doesnt remind me of anything just dull im sorry for the harshness im working in the field and see logos make or break brands and its usually an issue of low prioritization which I cant stretch how wrong it is
What is the point of having a togglable "pro-mode" as a free user if every simple search I use automatically uses my limited pro searches.
Please make it be how it was before
remove the toggle and make it a button and don't use your pro search if you have it deselected
this is the only reason I barely ever use perplexity
I’ve noticed that the shortcuts feature on Comet currently appears dull and lacks visual appeal, as there are no icons associated with the shortcuts. Additionally, the instructions are visible within the widget, which makes it feel cluttered.
I suggest improving this by allowing users to simply add a URL, similar to how Chrome shortcuts work, where the shortcut can open directly without additional instructions showing. Including icons for each shortcut would make the interface look cleaner, more engaging, and visually appealing.
Is the toggle not a button?
Oh I get it I think
Hi perplexity team.
Im a subscribed user
I couldn't find the support email (clicking Support button in perplexity won't work)
I'm testing Schedule task with Notification (In-app/Mobile/Email)
However Email notification won't work (I tried several times)
Could you please help me with this ?
Thanks
Hello! I very like perplexity, but I would work a little bit on visualisation and comfort for users. I would put on the general menu all chats or projects for faster interaction. Also as a user I would like to change chat names, now you can not do it. Please 🥺, can you add this feature?
BEHOLD, markdown headers
use that to enhance your suggestion into a readable list please
Want to inject some flavor into your everyday text chat? You're in luck! Discord uses Markdown, a simple plain text formatting system that'll help you make your sentences stand out. Here's how to d...
I'd also like to reiterate my earlier suggestion : make Comet run model(s) locally
- offloads some charge on your servers
- gives justification to all the AI hardware manufacturers sold to consumers for the past 5 years
-# nobody uses NPUs because no one bothers making useful apps for them, AI accelerators in GPUs are only ever used for upscaling - do it before someone starts competing with you with their own browser that runs models locally
- it'll be more secure than the current always-online solution
and a lot of the AI sceptics as myself will try it out on their own, see how it runs on various kinds of hardware and likely share about it with others (think Level 1 Tech, Jeff Geerling and many other techtubers)
I personally refuse to use something that doesn't run on my PC and stays there.
I’d like to suggest adding a feature that allows users to mute audio directly from the tab context menu. For example, when right-clicking on a tab, there should be an option to 'Mute Site' or 'Mute Tab.' This would improve usability by giving users quick control over unwanted audio without needing to switch tabs or adjust system volume.
almost everytime i ask perplexity to find evidence of something adn then quote it, it hallucinates
Just to vent, the Notion connector is pathetically bad. It'll connect 1 time, then it goes full on hallucination, no longer accessing Notion despite having the access.
Comet for Linux
Is it a safety issue that comet stores my passwords, and can use them to log in to websites when I ask it to do a task for me?
Yeah, it’s crazy wasn’t released alongside mac and windows
add a button to switch between accounts youre logged into instead signing out and signing in again
Hey it'd be great if we could get larger text in the chat & live assistant mode with a bigger chat box.
It would help out the vision impaired and blind app users.
Like having an accessibility setting?
I just started using Comet, and I love the idea of spaces, however it would be great if we could pin specific links, sites, as widgets to spaces instead of just saving threads
Congratulations on sponsored content.
Feedback: Turbo Tax sponsored ads in a response to Probiotics query isn’t the best use of the slot. A localized or targeted advertisement around gut health or longevity would be better place.
To Developers: I would like to Proposing improvements; Full chat read aloud, adjustable text size in the chat menu and Adjustable Text Size for the Live Speaking Mode. For Mobile & Tablet Users. Aimed at a clean interface encouraging thoughtful prompt implementation
↑Dont forget↑ this would be good for vision impaired & Blind for making everything more accessible
Customers paying $20-$100 per month shouldn't get obstructive ads for your products. (I know it can be fixed with uBlock but still)
Edit: The full screen popup is shown every time when opening the main site. Also upon pressing cmd+K for new conversation.
Legal documents can't be reached, like https://www.perplexity.ai/nl/hub/legal/terms-of-service
I need those to get the tool approved at work
Thank you for the link, but I still have the same issue:
(Page not found, code 404, check the link and try again, go back home)
No Bookmarks export option on Comet, as well as any other sync option or conformation of auto-sync. There was a password export, which is that. I'm resseting my PC, I wish preferences and bookmarks would sync, I can only home (searched sttings, online, nothing,)
Perplexity’s voice-to-text keeps repeating my sentences, making it hard to get a clear transcript. This bug really affects usage.
Is there a fix coming? Thanks!
any fix for this?
Add a podcast feature to spaces similar to NotebookLM.
Hi Team, for the past 4 days, Perplexity screen freezes after I typed the first three letters on iPhone 16pro/ios26.
I updated the app, deleted the app and installed it again, nothing changed.
What do you suggest?
Please allow me to change the name of the thread.
The iOS push notifications can only be turned on or off, no other settings. I want to be able to be notified of tasks that I create, that is the primary value I see in that feature. I do not want to be notified about events or whatever in the discover tab. The discover tab is completely accessory to the purpose of this app and I would happily ignore it accept that it conflicts with the primary functionality of a unique feature.
OMG Thank you sooooooooooooo much
Is it just me, or has Perplexity’s reasoning gotten noticeably worse since chat gpt 5 update with newer AI models?
Have you removed the ability to make HTTP REST calls from with Perplexity?
I have a number of projects that rely on this to make calls to APIs to get the most recent/accurate data.
For example - I have developed a stock market analysis model that relies on up to date market data and news that cannot be accessed from standard searches. Without getting data direct from APIs, perplexity frequently substitutes old or incorrect data - sometimes just outright lying. This is a big problem now for many of my workflows
I am using Perplexity Pro. Am I allowed to post images generated with it on Instagram?
it would be nice if perplexity can narrate a text you just wrote
Annoyance and I've reported it with no response from Perplexity team or Comet team, seems like gray area, but this promotional popup is an annoyance and interference to workflow.
Issue at hand, this pops up even after being dismissed within one sessions or within a short time interval, these are event controller on popup package, someone needs to fix this.
Love this app. i used GPT for so long and this by far is the best. I would like to see a public stock offering or possible integration with BBAI. I think Perplexity should be mentioned with the likes of GPT and claude and Grok!!
WHY IS IT THAT I SET MY SEARCH TO NOT USE PRO SEARCH AND IT STILL USES IT
THERE IS LITERALLY NO POINT OF THERE BEING A TOGGLE AT ALL??!?!?!!!
this has been suggested again and again and it seems like whoever is in charge does not give a toss about it
back in april i was considering paying for pro with the sale and potentially continuing, but forcing me to use pro just makes me not want to pay, at all
people keep saying to make the toggle-able pro search for free users actually work
yet it just doesn't happen
it used to work, where it could be toggleable where possible, but now i get pro searches for the simplest things, which if anything makes me think one of two things:
- perplexity is rubbish at figuring out what determines a pro search, making me not want to pay
- perplexity wants to push me to pay for "more pro searches", making me not want to pay
please let free users have a choice
i've had my perspective tainted by this tomfoolery, but that doesn't mean future users have to be disuaded because of this - given the previous feedback i'm clearly not the only one who thinks this, so don't keep making the same mistake
How convenient would it be if you could change it without having to enter the thread?
Extend the discord quest please 🙂
Hey team! 👋
I’ve been using Comet via my Perplexity Pro account and was loving the access to paywalled content on Le Monde and Le Figaro — worked great last weekend.
But now it seems that access is gone. 😕
Anyone else noticing the same? I thought Perplexity had deals in place with those publishers, so a bit surprised. Any idea what’s going on?
Thanks!
comet browser is very nice love it
Please add the function of bulk-editing queries which facilitate organising them into spaces or deleting
why can't I call perplexity assistant on my phone like I can call google assistant on my phone with hey google
I was deep diving in to a discussion on Chiropractic treatments and risks there of and suddenly I get "sorry I can't help with that!" To many restrictions seems to have put on the service again. And to make it discuss about 9/11 and how tragic plus my own memories of the event I had to say September 2001 to get the conversation going as "9/11" apparently triggered a restriction. 😩
In Comet the passwords viewer has a bug
🙂
Why can't Perplexity count emails in my connected Gmail account? I need to find invites and rejections and it just hallucinates
How can I join campus partner discord server
Hello everyone how to make perplexity labs project to last longer/ do more research?
Why have I been getting the "Upload failed" notification so often when uploading attachments lately?
I think wrong format or you didn’t wait long enough or maybe network
Can we make the answer instructions for Spaces actually effective? There's no point to putting in effort to create them and constantly remind the AI to adhere to them if it's going to immediately disregard them. Without fail, the AI will immediately forget about any answer instructions to follow.
I’ve been uploading .CS files, and even though it’s the same file, it keeps showing “Upload failed” every time I try to upload it.
it would be nice to be able to resume voice mode sessions, or start one in an existing thread
hi
I’m not a mod or anything just a normal civilian so idk🥲
perplexity need to launch a cli terminal for its own ai to do daily task and speed up process! otherwise why pay for perplexity if delaying many other stuff form other companies
It would be excellent, if there is an option to choose what AI to use. Instead of already being preloaded with GPT, Grok. I don't know, maybe some like Mistral, some Hugging Face and what was it, who knows, ah, duck.ai. I sure I forgot something.
Is Perplexity also in Visual Studio Code? That would be exccellent. Gotta look if that is.
Pls add option to use several gmail accounts for Max assistant and in general connection to search.
I have code for 1 year pro plan but i dont know how to activate it please help me
please contact support@perplexity.ai
Why is the client setup so you can not go in a out of voice mode? Ut really limits the utility
Please make Sync available in comet browser. Getting ready to switch back to chrome, since they now have all the features comet does.
Hey there dev team, i want to know if there is a project to make perplexity and comet compatible with linux based systems, it would be awesome
Will be there Turkish language support for both Comet and Perplexity? It’s been very long time it has not been added
Hello guys sorry for bothering you but does anyone knows why I can't choose the ai model i want while my ios friend can choose whatever ai model he wants? And we both have the pro version
Hi, there I would love to report the problem with the AI, as i asked him like 4 questions 3 are wrong, it is the first time me using it
For example I asked him to give me a sentence off of the book which the AI gave me a sentence that does not even exist
I have been using Perplexity and the Comet Browser for 14 days now and I am very satisfied. I am studying for my IT specialist retraining with Perplexity. Perplexity explains everything very well and even creates a knowledge quiz related to the topic.
Hi, could you please prioritize a proper start page with real site cards/Speed Dial for the Comet Browser? Currently, the home screen only shows a tiny set of “frequently used” sites, which is extremely limiting for daily work and muscle-memory navigation. This is critical and frankly very inconvenient: without customizable tiles, folders/groups, drag-and-drop placement, resizable cards, and reliable sync, it’s hard to operate efficiently 🫠
HELP! SCREEN FREEZING!
Hi Team, for the past 7 days, Perplexity screen freezes after I typed the first three letters on iPhone 16pro/ios26.
I updated the app, deleted the app and installed it again, nothing changed.
What do you suggest?
So, its still refusing to be my girlfriend (very bad service on this topic) but the rest is pretty, except for
- the issue that the browser sometimes closes itself when the agent is working on a website
- the sound cant be disabled on a tab, it just continues when the song starts over again
- the profiles are okayish, yet when a profile is closed, i want exact the same chat i had before and the same pages open
Hi currently I cannot login with email address and code , do we know if there is a problem
Suggestions for future updates:
Improve browser performance and resource usage (RAM/CPU)
Auto-suspend inactive tabs for better efficiency
Make the UI smoother and lighter
Clearly show when the AI acts automatically, fix vulnerabilities like “CometJacking” and allow detailed permissions management
Add tab groups, labels, and personalized user profiles
Enable multitasking layouts (split-view/interface)
Smarter, safer autopilot mode; add voice commands; use context from multiple tabs simultaneously
Full support for Chrome extensions, integrations with productivity apps, improved mobile and Linux versions
More visual customization options and themes
Performance/dashboard panel, usage-based smart suggestions, and easy built-in feedback tools
“Auditable AI mode” (see AI history/actions), “Private AI mode” (no history), and clean reader for long articles
comet using wayyy to much for just 2 blank tabs
One thing I'd love to see in the PDF / Markdown exports is the option to include the prompts as well (i.e. the full conversation) as opposed to only the responses from Perplexity... Don't see a way to accomplish that at present.
Close all of them and reopen
How can I get support on billing? I created an account off the Paypal promotion and it's not appearing. I went around a few rounds of messages with automated support Sam and finally got it to say they would have a human contact me, but no one ever reached out...
drm
precisely 💀
email support@perplexity.ai and specify that you want human support
is Alex CEO? OWner? Whatever
Just curious
Would be great if CEO would actually engage in Discord with community
I signed up with a student account.
After 1 month of free use, I received an email saying there’s a $4.99 USD discount.
Will I pay $4.99 USD per month, or will I pay $15 USD with a $4.99 discount?
ask perplexity
how?
Question about maps in iOS. The app has shown me maps before, but this new space I created, specifically for travel and when I specifically asked the space to show me maps when possible… it’s getting confused and tries to create a chart with plotted dots of locations. Anyone else having trouble with maps? Not sure if it’s a bug or not
then ask copilot, or gemini
or chatGPT for that matter
one of them must know the answer
/@olive grove must know
hey it's gonna be nice if there's a listen mode like in the phone app but on web app
I am not able to click on links in iOS app. Anybody else facing same issue?
Friend got a link to get Perplexity pro
he tried everything to get it actually put Perplexity used SheerID which SUCKS
Guys, labs resets monthly?
Tried everything, but sheerid always says cant verify (they just ask for name of school, name, DOB and email)
He tried with personal email, school email etc etc and everything sheerid fails
Tried support.. never gets a reply
SheerID is really bad esp for people outside USA. Just read its reviews on reddit etc.. why can't perplexity just use student email OR another better partner? Or USE their OWN AI to figure it out??
in dark mode when we create api group the dropdown menu for the country and other's are not visually appealing for them the background color merging with the text
I invited my friend to comet but none of us received any award. He has already asked a question. Why is that?
yes
Could it be possible to set up custom rules/guidelines for the voice chat? There's just so much yada-yada. I would like to customize the behavior.
I think so yes
Titular con Noticias Importantes son necesarios como la paz en Gaza sin sesgo político de extrema izquierda. Perflexity no lo informó como merecía con titular por ser Trump! Inaceptable!
Don't do politics here
I'd like to reiterate an earlier suggestion : make Comet run model(s) locally
- offloads some charge on your servers
- gives justification to all the AI hardware manufacturers sold to consumers for the past 5 years
-# nobody uses NPUs because no one bothers making useful apps for them, nor for the AI accelerators in GPUs which are only ever used for upscaling - do it before someone starts competing with you with their own browser that runs models locally
- it'll be more secure than the current always-online solution
and a lot of the AI sceptics as myself will try it out on their own, see how it runs on various kinds of hardware and likely share about it with others (think Level 1 Tech, Jeff Geerling and many other techtubers)
I personally refuse to use something that doesn't run on my PC and stays there.
You can share this thread for the team to look at
The new IOS update breaks some things
Has anyone seen any feedback on if they will at some point allow multiple gmail connectors?
Please post it here #1393681130842226698
hey Perplexity team, I download the comet browser but even after its not getting installed on my laptop. I have tried multiple times but uable to do so.
Please share any screenshots or error messages that you are getting.
Try out these installers instead
Windows: https://www.perplexity.ai/rest/browser/download?channel=stable&platform=win_x64
MacOS: https://www.perplexity.ai/rest/browser/download?channel=stable&platform=mac_arm64
It would be extremely valuable if comet included a built‑in AI‑powered search and summarization tool. Instead of only finding exact keywords with CTRL+F, this feature would let users search pages, PDFs, and documents using natural language queries (e.g., ‘Where does this article explain the causes of inflation?’). It could also instantly summarize long documents and organize highlights across multiple tabs. This would transform comet into a true productivity and research platform, making it far more powerful for students, professionals, and everyday readers.
hello I've sent a support ticket for my billing issues since Thursday last week, i've had no response. can someone please help me? I'm still waiting for a human response.
Do I really have to see those pronunciations? How do I disable and just see the bare text
looking to get approved for the DUB perplexity program - stollthbel@gmail.com thank you!
@orchid pebble I think I found a fake Comet app on iOS App Store: https://apps.apple.com/hr/app/cornet-ai-smart-ai-browser/id6753550490
A fast, secure Comet AI browser with PDF conversion, AI chat, and private browsing for iPhone and iPad.
Minimalist Style: "Clean, minimal Comet AI browser for distraction-free browsing with built-in PDF tools."
Speed & Performance Focused: "Lightning-fast AI browsing, instant PDF conversion, and s…
Ty, I will forward this to Perplexity staff.
I was using Pro and upgraded to MAX, and I am no happy with the results in either. I am trying to use it to generate an 3D image and 3D video. The first instruction usually goes well. I am an engineer and used to writing specifications. I cab be very concise and precise in my wording. The problem start then. The generated content will be 90% correct, so I will be specific and say "Do not revise the uamge other tan to..." and in most cases it adds or revises what I asked PLUS changes something else not related, or if I try again I often get a TOTALY different content labeled "We kept the image the same changing only". Giving and extreme, but not uncommon example, see the two images the starting point and then the revision. The instruction was "2 Leaving everythig else the same add an accomdation module for 20 persons to the top deck." This would onl be frustrating, but then I hve twice run into "you have used your limiit of 15 images/videos this month". True but only because it take 10 or 20 tries to get it to do as asked. (You dont want to know how many wired results I got making image 1!!) Anyone else have this problem? And where does the 15 limit come in. Dont see it anywhere in my account info.
while using comet I asked perplexity to figure out why I had $0 on my dub account after sharing my given link to my friends.
I'VE PROVIDED A FRIEND WITH THE REFERRAL LINK. THE USED THE LINK, DOWNLOADED COMET, MADE IT THEIR DEFAULT BROWSER AND MADE A QUERY ON COMET; SO WHY IS MY DUB NOT SHOWING ANY MONEY EARNED?
this image is a lie, unfortunately i didnt find this out till i signed up for dub and stripe and gave them my information
this is how perplexity responded to my complaint: Based on the screenshot and the official Perplexity Refer-A-Friend webpage, there is actually no mention of a “student” requirement for this referral program. The terms simply state that:
- You and your friend must be eligible (over 18, with no previous Comet accounts for the friend, and based in the United States)
- Your friend needs to download Comet through your link, log into Perplexity, and ask their first question
No student restriction is mentioned anywhere in this offer or the program terms for the $15 referral reward. The conditions are about eligibility, device, and using the correct referral link—not your student status or country (except that the offer is only valid in the US).
If anyone—regardless of student status—fulfilled these steps and your reward did not show up, Perplexity and Dub support owe you a clear answer as to why you didn't qualify.
You are correct—there was no way to know from the promo image or the site that it was student-only, or that you risked sharing personal data for nothing. If you would like, you can escalate this to Perplexity or Dub support and reference their published terms for further review.
so if someone from perplexity/comet could do the right thing and give me a clear reason for why my efforts have gone unrewarded that would be almost decent of you
@chilly dagger the rewards depends on the country they are in, it s different for most of the countries and the reward rn may be bugged because i invited 5 friends and got 10$ for each, (50$ in total) and for the 6th and 7th i didn t get the reward, and perplexity team said they will solve the issue. you should try to wait a bit or just tell them to solve the issue.
sorry if i have typos, im not a native english speaker
can someone explain this to me? i asked for the cube root, as this solution is incorrectly written
Is it possible to see the previous versions of an edited prompt like you can in ChatGPT?
I had been enjoying building useful apps in Perplexity labs, however as they have recently forced the app to run in a temporary pop-up browser it makes it almost unusable. It seems aimed at checking code, not running something in production. I don’t know where they’re going with this feature, but it makes me hesitant to continue to invest time on that platform.
hey Perplexity team, I download the comet browser but even after its not getting installed on my laptop. I have tried multiple times but uable to do so.
Can someone help me
I have airtel plan active and my subscription got cancelled yesterday night
please contact support@perplexity.ai
I have but no reply
No image is generated—even though I have not exceeded any daily limit and everything is set up correctly. Is there any server-side issue or technical bug that the team is aware of? If yes, please advise on how soon image generation will be restored. If you need screenshots, I can provide them.
Hello, my name is Evgeny, I'm a new user from Russia!
About the Comet browser:
It would be really convenient if the browser could open desktop apps via links, for example Steam or Discord – so when I click the link, it opens the actual desktop application, not the web version.
Please add alternative voice recognition engines and enable seamless communication for Russian-speaking users – both voice input and output should work perfectly in Russian.
I'd love to see more customization options for the interface, such as advanced theme settings, color schemes, and layout personalization.
Overall, I really like the browser! Thank you, Perplexity team!
Android Perplexity:
I'm quite happy – it works quickly, but the design feels a bit underwhelming compared to the beautiful web version. Thank you!
Desktop Perplexity:
I wish it were not just a web app, but a full-fledged assistant able to control my PC and perform system actions – like a smart desktop assistant for Windows. Thank you!
Regarding Perplexity in general:
I would really appreciate the ability to verify my age and access all neural network features without limit – generate any images and texts I need, even if they formally violate the current policy (please consider revising the rules). I really enjoy your app, but these restrictions are quite annoying! Thank you.
-Translated and improved with the help of Perplexity :D-
i dont how people find perplexity find useful at all i dont,
just being literal.. like not once or twice, it always is gonna give a 10 page ans nothing less
isnt that what you want? its giving more information
guys fix the issue where sometimes perpleixty assistant is blocked on some websitres
yeah mine isn't either lol
and I literally referred all my college friends which are studnets
@covert sun @native tartan
@native tartan
also lol
I applied for the perplexity program
u really think i want that kind of more information!!??
copilot summarises more than perplexity.. i would rather go myself and click on blue links than to find the valuable part out of perplexity ans
Hi Team,
Not sure if this is a right channel to ask, today I noticed that perplexity ai is giving free pro to the students, I registered myself and got the approval. However, it's still asking me to pay $4.99 per month and showing a $15 discount.
Also, if I switch with annual it's again showing a $15 discount only.
Could you help me with this - how do I get free pro?
Thanks,
You and your friend must be eligible (over 18, with no previous Comet accounts for the friend, and based in the United States)
Hello. I just added the Perplexity bot to my Discord server, but it’s not responding to my commands. How should I set this up?
Defining shortcuts with / does not respect the disabled web search
Has a usage limit for advanced models been added to the search mode of the Pro plan?
I was using it just now when it displayed a message saying I had reached the usage limit.
Learn more about usage limits here - #💬│general message
Because I've been on Comet for seven hours trying to do something and I'm starting to have a nervous breakdown. The thing crashes every time, I don't know what to do anymore, I've wasted seven hours of my time when the basic tool is supposed to save me time, so I've decided to go for the ZERO version...
In short, I think I'm going to unsubscribe and get my money back because it's really not up to scratch! And since no one seems to take into account that when you sell on the French market, the minimum requirement is to provide French support...
I cannot believe the rapid advancements the dev team has made over the last 1-2 years. Keep killing it.
what kind of a silly statement is this
…me saying that I’m impressed with Perplexity compared to other LLM’s.
What kind of silly question is this? @ocean jacinth
rapid advancements 🙋
Yes. Last I checked 2 years ago it was nothing like this?
long 2 years .
and don't bother praising anyone here ,it's pointless
Perplexity's ability to store the context of the OP question has completely gone to shit over the last 2 months IME. I find myself constantly having to open new threads because it can't follow along to the basic topic in the OP question anymore. I barely ever had this problem previously and now it seems to happen like 30% of the time. Latest example, I asked it to summarize a news article on (former) Prince Andrew, fine, then I ask a question about the article "what were the Chinese spy allegations" and it just talks about Chinese Spy allegations in the general sense, when obviously I wanted to know about the allegations against Andrew. Previously it was smart enough to recognize this, I don't know what happened. Thanks for reading, hope someone can improve this feature, I used to like using perplexity a lot.
Can someone help me
I have airtel plan active and my subscription got cancelled yesterday night
I am in china, why I can not access to perplexity, ?
would be nice if you can use perplexity to target the content in a given space. for example, summarizing the information across multiple conversations
Add the ability to change your account email address
Use vpn
please contact support@perplexity.ai
guys is there a way to stop websites from blocking comet automation
it happens too regularly
Please fix this. If I choose the LLM option (which I think is perfect for me), and then I ask questions like “teach me” or “provide me something,” it automatically switches to the preset “study” section. But it shouldn’t do that unless I choose it manually.