📢announcements

[bleak hinge]

We have a dedicated channel for self-promotion: #✨projects

Please limit all the promotion to that channel only, messaging individual users to advertise your product will get you banned.

To users: Report such spammers by sending the screenshot of their messages via #❓support . Thank you.

[bleak hinge]

@everyone Thanks to @proven gyro, we now have @wintry knoll on the server. It's a GPT Bot that answers questions using GPT-3, written by Jer. For help type “!h help”.

**To ask a question type “!q YourQuestionHere” ** or just ping the bot or use /question command

Example: !q Can you tell me a story about a bot named ChatGPT?

Use it in only #🤖gpt-bot-1 and #🤖gpt-bot-2

[bleak hinge]

From now on, messages with 10 stars will be featured in #⭐starboard , you can react with a star ⭐ to helpful tips, or any interesting thing. Folks featured on #⭐starboard will get a custom role.

[oblique pasture]

Hey @everyone, another big announcement:

ChatGPT Discord Bot
Thanks to @vocal jackal, we now have the highest-quality GPT chatbot on Discord. We're excited to announce the latest addition to our bot family - @dense ocean! ChatGPT is actually using ChatGPT - not the OpenAI API, but interacting directly with https://chat.openai.com/chat.

Features: Individually stored conversations for each user, the ability to /reset your conversation thread.

The bot is still under active development so will likely see some instability and changes, especially while the developer goes through the process of acquiring more accounts to keep up with future demand.

Chat with ChatGPT by using @dense ocean in the following channels:
・ #New Channel Name
・ #New Channel Name

Also remember the #🪧rules! Any attempts to use the bot to bypass/break them will result in timeout and/or ban!

[bleak hinge]

@everyone Thanks to https://www.bluewillow.ai/

we now have @raw pumice on the server. It's an AI image generator bot and the best part? It's free and has unlimited uses! It currently serves thousands of users so please be patient if it doesn't complete your request immediately.

How to use the bot:

1. Go to #🎨image-ai or #🎨imagebot-2

2. Submit a prompt:
   Type /imagine prompt: or select the slash command /imagine from the pop-up menu.

3)Describe in the prompt box with text what image you would like to generate.

4. Send the command!

🇺: The function of "U buttons" is to enhance the visual quality of an image by creating a larger, more detailed version of the selected image.

🇻: The function of "V buttons" is to produce alternative versions of a selected grid image. The result is a new grid image that retains the original's overall aesthetic and composition, with subtle modifications.

🔄: It restarts a generation by generating a new grid of images based on the original prompt.

⚠️Note: Server rules still apply. NSFW requests will not only get you banned from the server but also from the bot developers.

https://media.discordapp.net/attachments/1072152113465004052/1072507585510506516/FAQs-07.jpg?width=566&height=566

[bleak hinge]

New bots addition:
@wintry knoll in #🦋adobe-firefly - Art Generator use /fiireflyimage

@minor osprey in #⏸labs-ai - TTS Voice cloning use /elevenvoice

[bleak hinge]

https://i.imgur.com/eUsldTV.mp4

[craggy zinc]

GPT 4-Turbo
• GPT 4-Turbo is now available via the API
• 128k token context (vs. 32k previously)
• More than 50% cheaper than GPT-4
• April 2023 Knowledge cutoff (vs. Sept 2021)
• Improved performance from GPT-4
• Can now provide images along side your text input prompt via API
• A new JSON mode that forces GPT to respond in pure JSON
• More generous rate limiting

Custom GPTs
• Users can build their own "Custom GPTs" that are specialized for specific tasks
• You can make CustomGPTs without coding. Use natural language, and upload files for context
• Enterprises can make company & org-specific Custom-GPTs
• OpenAI is giving us two Custom GPTs to start: Canva and ZapierAI

Custom GPT Store
• Users can upload their CustomGPTs to the store for others to use
• OpenAI will provide a revenue share program, where authors of popular models will receive $.

Assistants API
• The assistants API lets you build autonomous agents that have access to tools.
• OpenAI currently provides 3 tools to start: Code Interpreter (coding), Retrieval (Custom Knowledge), and Function Calling
• Just like with the normal API, you specify its role via a a custom instruction.

Human Quality Speech
• Open AI released the tts-1 and tts-1-hd models.
• The tts-1 model is optimized for speed, and tts-1-hd model for quality
• Choose from 1 out of 6 voice types and create realistic sounding human speech from an API

Copyright Shield
• Protects you and your company from copyright claims when using OpenAI's products

Whisper V3
• As a quick recap, Whisper is OpenAI's speech-to-text model. It transcribes voice and outputs text.
• Whisper is open source, and V3 was also released as open source.
• It seems via the API (paid), Whisper v3 is not available yet at the moment.

Custom Models for Enterprise
• For select companies, OpenAI's research team will create a custom model for your enterprise with specific domain knowledge.

[craggy zinc]

Happy birthday to ChatGPT!

[craggy zinc]

Happy birthday to my new daughter!

[craggy zinc]

Happy new year everyone! It's been a fun one. What do you think AI will bring us in the next year?

Share your predictions

[bleak hinge]

Hey @everyone — we're looking for subreddit and discord mods! Apply here:

Subreddit

Discord

[craggy zinc]

Review completed & Altman, Brockman to continue to lead OpenAI

New board members named and enhancements to the governance structure introduced

• The OpenAI Board has completed a review, conducted by WilmerHale, evaluating the November 2023 removal of Sam Altman and Greg Brockman from the OpenAI Board and Mr. Altman’s termination as CEO.
• The review found a breakdown in trust led to their removal, but not due to concerns about product safety, company finances, or development pace.
• Following the review, the Board expressed full confidence in Altman and Brockman, reinstating Altman as CEO and both as leaders of OpenAI.
• Three new Board members were elected: Dr. Sue Desmond-Hellmann, Nicole Seligman, and Fidji Simo, to support and oversee OpenAI's growth.
• OpenAI announced enhancements to its governance structure, including new corporate guidelines, a strengthened Conflict of Interest Policy, a whistleblower hotline, and additional Board committees.

Read more on OpenAI's blog

[craggy zinc]

OpenAI "accidentally" published and quickly pulled a blog post about GPT-4.5 turbo

[craggy zinc]

New update to GPT-4-turbo

[craggy zinc]

https://platform.openai.com/docs/api-reference/batch
OpenAI has made a new API update for developers, allowing them to save 50% on API costs for tasks which aren't speed-critical. Send your prompts in in a large batch and receive the results in <24 hours.

[craggy zinc]

🎉 🥳
Happy 39th birthday Sam Altman

[craggy zinc]

Moderna has partnered with OpenAI using ChatGPT Enterprise
https://www.youtube.com/watch?v=t3UHnKLVS1M

[craggy zinc]

Stack Overflow and OpenAI have partnered to create better LLMs
https://stackoverflow.co/company/press/archive/openai-partnership

[bleak hinge]

🚨🚨 @everyone We are hosting an Ask-Me-Anything (AMA) session with OpenAI CEO Sam Altman 🚨🚨

Come join us on Reddit and ask all the questions you have: https://www.reddit.com/r/ChatGPT/comments/1coumbd/rchatgpt_is_hosting_a_qa_with_openais_ceo_sam/

[craggy zinc]

Missed the OpenAI spring update stream? You can watch it here if so!

Here's a summary of the new changes:

• Releasing the desktop version of ChatGPT to make it more accessible and integrated into users' workflows
• Refreshed the ChatGPT UI to make the interaction more natural and intuitive
• Launching a new flagship model called GPT4O that brings GPT-4 level intelligence to all users, including free users
• GPT4O has improved capabilities across text, vision, and audio, with real-time responsiveness and the ability to perceive and respond to emotion
• Bringing GPT4O to the API so developers can build advanced AI applications with it
• Expanding the GPT Store to allow more creators to build custom ChatGPT models for specific use cases, making them available to a broader audience
• Adding new capabilities like vision integration, memory, and advanced data analysis to ChatGPT
• Improving quality and speed of ChatGPT in over 50 languages

[craggy zinc]

https://openai.com/index/hello-gpt-4o/
It seems that OpenAI also updated DALL-E! New image generation seems significantly more coherent than before.

[craggy zinc]

Missed the Google I/O 2024 stream? You can watch it here if so!

Here's a summary of the stream:
Gemini:

• Introduced Gemini 1.5 Pro with a 1 million token long context window, now available globally. (Available)
• Expansion of context window to 2 million tokens for developers in private preview. (Future)
• AI Overviews in Search rolling out in the US this week. (Available)
• Ask Photos feature in Google Photos launching this summer. (Future)

Google Workspace:

• Rolling out Gemini 1.5 Pro in Workspace Labs with improved email summarization, and meeting recap capabilities. (Available)
• Prototyping new virtual teammates using Gemini for specific tasks. (Future)

NotebookLM:

• Adding Gemini 1.5 Pro capabilities, including personalized science discussions with audio overviews. (Available)

Generative Media Tools:

• Introduction of Imagen 3 for improved photorealistic image generation. (Available)
• Introduction of Vio for high-quality 1080p video generation, releasing to select creators soon. (Future)

AI Agents:

• Prototypes for intelligent AI agents for tasks like shopping and organizing tasks; deeper integration planned. (Future)

New Hardware:

• Introduction of sixth-generation TPUs (Trillium) for better AI model performance, available to cloud customers in late 2024. (Future)
• Upcoming availability of NVIDIA's Blackwell GPUs in early 2025. (Future)

Search Innovations:

• Multi-step reasoning in Google Search for complex questions and planning tasks. (Different features rolling out soon)
• AI-organized search results page for brainstorming and inspiration tasks, starting with dining and recipes. (Future)
• Adding video understanding to Google Search. (Future)

Gemini as an Android Assistant:

• Gemini becoming more context-aware and integrated into Android for more helpful tasks. (Available)
• On-device AI with Gemini Nano for speed and privacy, launching on Pixel later this year. (Future)
• AI-powered fraud detection capabilities in Android under testing. (Future)

Gemma Models:

• Debut of PolyGemma, an open vision language model available today. (Available)
• Upcoming launch of Gemma 2 in June with a new 27B parameter model. (Future)

Responsible AI Initiatives:

• Expansion of SynthID for text and video watermarking to prevent misuse. (Future)
• Open-sourcing SynthID text watermarking and providing it in the Responsible Generative AI Toolkit. (Future)

Learn L.M.:

• Introduction of Learn L.M. for educational purposes, incorporating it into products like Search, Android, Gemini, and YouTube. (Rolling out soon)
• Learn L.M.-powered Learning Coach Gem launching in the Gemini app in coming months. (Future)

[craggy zinc]

OpenAI has partnered with Reddit
https://openai.com/index/openai-and-reddit-partnership/

[craggy zinc]

https://openai.com/index/openai-safety-update/

[craggy zinc]

OpenAI has recently begun training its next frontier model and we anticipate the resulting systems to bring us to the next level of capabilities on our path to AGI.
https://openai.com/index/openai-board-forms-safety-and-security-committee/

[craggy zinc]

OpenAI has announced several bans on government actors using its services to covertly influence social media
https://openai.com/index/disrupting-deceptive-uses-of-AI-by-covert-influence-operations/

[craggy zinc]

New promo video for the new voice mode
https://www.youtube.com/watch?v=4w0Pqs3CuWk

[craggy zinc]

Discord has announced an update to their developer policy. This new update now restricts the use of AI tools on Discord user data, unless express permission is granted by Discord staff.

16. Do not use API Data to: profile Discord users, their identities, or their relationships with other users...
17. Do not use message content obtained through the APIs to train machine learning or AI models (including large language models) unless express permission is granted by Discord.

Read more here:
https://discord.com/developers/docs/policies-and-agreements/developer-policy

[craggy zinc]

New OpenAI premiere, promoting the advanced voice mode
https://www.youtube.com/watch?v=Mckd-FhJlp0

[craggy zinc]

OpenAI has announced a new partnership with the Los Alamos National Laboratory. OpenAI will be assisting with biological research
https://openai.com/index/openai-and-los-alamos-national-laboratory-work-together/

[craggy zinc]

If you'd like, contact @grim grotto with questions about submitting your startup.

[craggy zinc]

Babe, wake up, new model just dropped
https://openai.com/index/gpt-4o-mini-advancing-cost-efficient-intelligence/

[craggy zinc]

https://openai.com/index/searchgpt-prototype/

[craggy zinc]

There's a new version of GPT-4o out which can output 64k tokens at once.
https://openai.com/gpt-4o-long-output/

[craggy zinc]

OpenAI has announced their 2024 Dev Day events. The events will be on the following days:

• San Francisco: October 1st
• London: October 30th
• Singapore: November 21st

We'll be restreaming the Dev Day conferences live here!
For more info, check OpenAI's website.

https://discord.gg/r-chatgpt-1050422060352024636?event=1270125893146382460

[craggy zinc]

OpenAI API has updated. This update allows much stronger guidance of JSON formatting of API responses
https://openai.com/index/introducing-structured-outputs-in-the-api/

[craggy zinc]

OpenAI has released the system card for GPT-4o, detailing their red teaming strategy and the associated risks
https://openai.com/index/gpt-4o-system-card/

[craggy zinc]

Strawberry has been confirmed and is being discussed officially now
https://x.com/polynoamial/status/1834275828697297021

Noam Brown (@polynoamial) on X

One year ago, I joined @OpenAI hoping to create AI models capable of truly general reasoning. Today, I’m excited to share with you the fruit of that effort and the hard work of many amazing teammates: OpenAI’s o1 model series! (aka 🍓)
https://t.co/inZfIzLIBB
Let me explain 🧵 1/

Twitter · Sep 12, 2024, 5:00 PM

lmfao he hit send too early. it's deleted now

OpenAI's newest model is dropping now!
https://openai.com/index/learning-to-reason-with-llms/
@everyone

[craggy zinc]

Don't have ChatGPT Plus? No problem. You can try o1-preview on @topaz stone in #🔥gpt-5 for free. Just make sure you set your model preference so you're getting responses from o1-preview instead of older models.

[craggy zinc]

According to an OpenAI employee, the full version of O1 will drop in about a month.
https://x.com/dmdohan/status/1834284226013003946

David Dohan (@dmdohan) on X

It's important to emphasize that this is a huge leap /and/ we're still at the start

Give o1-preview a try, we think you'll like it.

And in a month, give o1 a try and see all the ways it has improved in such a short time

And expect that to keep happening

Twitter · Sep 12, 2024, 5:33 PM

[craggy zinc]

Sam Altman departs OpenAI's safety committee
https://techcrunch.com/2024/09/16/sam-altman-departs-openais-safety-committee/

[craggy zinc]

Only 11 more days til Dev Day. 🔥
https://discord.gg/r-chatgpt-1050422060352024636?event=1270125893146382460

[craggy zinc]

https://x.com/sama/status/1838864011321872407
Advanced voice mode rollout for Plus users is complete

Sam Altman (@sama) on X

rollout completed early, amazing work by the team!

(except for jurisdictions that require additional external review)

Twitter · Sep 25, 2024, 8:51 AM

[craggy zinc]

OpenAI now has a similar system to Anthropic's "artifacts" system. It's called canvas and you can use it now.
https://openai.com/index/introducing-canvas/

[craggy zinc]

Anthropic has announced a new version of Claude 3.5 Sonnet, along with announcing the release of Claude 3.5 Haiku. They have also announced a new interface for Claude to interact with your computer directly, as a human would.
Claude 3.5 Haiku and Claude 3.5 Opus aren't yet released, but the new version of Sonnet is. See the attached photo for benchmark info.
https://www.anthropic.com/news/3-5-models-and-computer-use

[craggy zinc]

Shortly after Anthropic's new version of Sonnet released, Google sent out an email to Google Cloud customers, promising heavier surveillance and making vague threats of account termination.
Solid move Google, thank you very much 🤡

[bleak hinge]

🚨AMA (Ask Me Anything ) with OpenAI’s Sam Altman, Kevin Weil, Srinivas Narayanan, and Mark Chen: https://www.reddit.com/r/ChatGPT/comments/1ggixzy/ama_with_openais_sam_altman_kevin_weil_srinivas/ @everyone

[craggy zinc]

Don't feel like reading the whole thread? Here's a summary based on the AMA hosted by the OpenAI team:

• New Features and Products:

  • The team confirmed that they are working on enhancements to ChatGPT, including longer context windows.
  • There are plans for future models but nothing specifically named as GPT-5 is expected in the near term.
  • Work is ongoing to decrease hallucinations and improve model accuracy.
  • Advanced Voice features are being developed, with aspirations like singing capabilities.
  • A new text-to-image model is in development, although they did not share a release date.
  • They hinted at improvements in search capabilities, noting the release of "SearchGPT."
  • Custom web page rendering in response to search queries is a potential future feature.
  • They mentioned the possibility of integrating vision for interacting with local environments.
  • Plans for integrating agents that can perform tasks autonomously by 2025 were discussed.

• Technical and Development Insights:

  • They are focusing on both scaling base models and improving compute time for faster inference.
  • There's an emphasis on ensuring AI safety and alignment as they develop more advanced models.
  • The team is exploring better personal information management and action-taking capabilities.
  • They acknowledged the challenge of potentially opening source some of their older models and software.

• General AI and Future Vision:

  • The OpenAI team re-affirmed their belief in the achievability of AGI (Artificial General Intelligence) with current hardware.
  • The discussion included the potential for AGI to accelerate scientific discovery, particularly in health and medicine.
  • They addressed the balance between openness and safety due to the risks associated with powerful AI technologies.
  • The team is considering how to incorporate human-like features in AI interactions, especially in Advanced Voice and potentially future models.

• Community and Social Implications:

  • There was a mention of potentially relaxing content policies for adult users, but with current priorities focusing elsewhere.
  • They're navigating how AI might impact job markets but expressed optimism about improving productivity and creating new opportunities.

• Miscellaneous:

  • The team shared personal insights and a bit of humor, like Sam Altman's preference for lowercase typing and naming a hypothetical OpenAI pet.
  • OpenAI is engaged in ongoing negotiations and considerations with international policies and regulations.

[craggy zinc]

Happy birthday ChatGPT! 🥳

[craggy zinc]

Happy first birthday ChatGPT! (to clear up any confusion, my daughter's legal name is ChatGPT)

[craggy zinc]

Twelve days of OpenAI releases starting tomorrow! 🥳 🎄
https://x.com/openai/status/1864328928267259941?s=46

OpenAI (@OpenAI) on X

12 days.
12 livestreams.
A bunch of new things, big and small.

12 Days of OpenAI starts tomorrow.

Twitter · Dec 4, 2024, 3:20 PM

[craggy zinc]

It looks like we'll be getting a "limited preview of GPT-4.5" as part of the 12 Days of Shipmas 🤯
Source

[craggy zinc]

🥳 The full version of OpenAI's o1 model has released, along with their new subscription plan which includes o1 Pro. Join us again tomorrow for the second day of Shipmas
https://openai.com/index/introducing-chatgpt-pro/

[craggy zinc]

LMArena has a new project, WebDev Arena. Describe the website you want to see, then vote on which model did a better job.
https://web.lmarena.ai/

[craggy zinc]

The limits on Sora have been lifted for the holiday season. Merry Christmas!

[craggy zinc]

Ho ho ho, Meeeeeeerry Christmas!

[craggy zinc]

2025 is right around the corner. Here's a list of the stuff people predicted in [last year's prediction thread](#1191170684181819523 message):

I predicted GPT-4.5 in March ❌, GPT-5 in late summer or early fall ❌, Google closing the gap on OpenAI especially in pricing ✅, slow progress in multimodality ❌, and open source models surpassing GPT-3.5 and GPT-4 ✅. 2/5, pretty bad

@shrewd crescent predicted GPT-5 in February. ❌ 0/1

@novel spade predicted Midjourney video generation. ✅ 1/1

@violet edge predicted open source reaching GPT-4 ✅ and a GPT-5 release ❌. 1/2, but not that far off being that we now have o1.

@flint epoch predicted Bard (now Gemini) integration into YouTube ✅ and predicted it would cause issues ✅. 2/2

@manic apex predicted ASI behind closed doors. ???/1

@jovial lodge predicted gains vs GPT-4 by open source✅, Llama 3 by July✅, efficient models and chips for running models locally✅, and ease of use for consumers making their own fine-tunes.✅ 4/4, excellent predictions.

@clear island predicted that GPT-5 would not release this year✅ and that new models would be price-prohibitive✅. 2/2

@strong burrow predicted GPT-5 in late Q3 or Q4 of 2024, possibly with an earlier test release. Though o1 isn't called GPT-5 this feels mostly correct. ✅ 1/1

@runic kindle and @deft tree predicted AI regulation would hinder progress ❌ and thankfully was wrong! 0/1

@red stirrup predicted improvements in video AI ✅. 1/1

@deft horizon predicted that death caused by AI would appear as a major headline ✅. Lawsuits against character AI are ongoing, so sadly, 1/1

@signal thicket predicted 1M context models ✅, refactoring and language translation improvements in AI✅ and hybrid models with advances in multimodality✅. 3/3, great job HorseCode

Make your 2025 AI predictions here!

[craggy zinc]

AMA on reddit with Sam Altman and other OpenAI staff

[craggy zinc]

https://www.youtube.com/watch?v=jv-lpIsnLOo

[bleak hinge]

https://www.reddit.com/r/ChatGPT/comments/1il23g4/calling_ai_researchers_startup_founders_to_join/

[craggy zinc]

News regarding GPT-4.5 and GPT-5 release

[craggy zinc]

https://www.theverge.com/notepad-microsoft-newsletter/616464/microsoft-prepares-for-openais-gpt-5-model

[craggy zinc]

https://www.anthropic.com/news/claude-3-7-sonnet

[craggy zinc]

Deep Research is no longer restricted only to Pro users. 🙂
https://fxtwitter.com/OpenAI/status/1894454194943529433

Claude 3.7 Sonnet is currently playing Pokemon live. You can watch it here:
https://m.twitch.tv/claudeplayspokemon

[craggy zinc]

OpenAI livestream today at 10 AM PT!
https://x.com/OpenAI/status/1899476049584599462

OpenAI (@OpenAI) on X

This one's for the devs.

Livestream at 10am PT.

X · Mar 11, 2025, 3:02 PM

[craggy zinc]

OpenAI livestream in 30 minutes!
https://www.youtube.com/watch?v=lXb0L16ISAc

[craggy zinc]

OpenAI livestream today. Looks like we're finally getting native image outputs!
https://x.com/OpenAI/status/1904556394847862809

OpenAI (@OpenAI) on X

X · Mar 25, 2025, 3:30 PM

[bleak hinge]

@everyone AMA with Head of Model Behaviour @ OpenAI is Live Now!

https://www.reddit.com/r/ChatGPT/comments/1kbjowz/ama_with_openais_joanne_jang_head_of_model/

[craggy zinc]

https://openai.com/index/sycophancy-in-gpt-4o/

[bleak hinge]

@everyone
AMA with OpenAI codex team: https://www.reddit.com/r/ChatGPT/comments/1ko3tp1/ama_with_openai_codex_team/

[craggy zinc]

https://www.youtube.com/live/o8NiE3XMPrM

[craggy zinc]

https://x.com/AnthropicAI/status/1925239440420831516

Anthropic (@AnthropicAI)

Join us from 9:30am PT tomorrow: https://t.co/3lSlYKqQCT.

X · May 21, 2025, 5:17 PM

[craggy zinc]

https://sean.heelan.io/2025/05/22/how-i-used-o3-to-find-cve-2025-37899-a-remote-zeroday-vulnerability-in-the-linux-kernels-smb-implementation/

[bleak hinge]

@everyone
Join in for the AMA with PerplexityAI Co-Founder Aravind Srinivas: https://www.reddit.com/r/ChatGPT/comments/1m1javp/comet_ama_with_perplexitys_aravind_srinivas_and/

[craggy zinc]

OpenAI has released two new open source model 🥳
https://openai.com/index/introducing-gpt-oss/

[craggy zinc]

https://x.com/openai/status/1953139020231569685

OpenAI (@OpenAI)

LIVE5TREAM THURSDAY 10AM PT

X · Aug 6, 2025, 5:00 PM

[craggy zinc]

lmfao

[craggy zinc]

Just welcomed my second child into the world! Meet Claude4 Opus Kelly! (If there are any Anthropic employees in the chat, please hit me up so I can get those free API credits)

[craggy zinc]

https://www.businessinsider.com/hunger-strike-anthropic-deepmind-ai-threat-2025-9
lmao

[craggy zinc]

https://x.com/MichaelTrazzi/status/1966285365758628240
doctors inform genius that not eating is bad for you, he quits

Michaël (in London) Trazzi (@MichaelTrazzi)

I've just received a medical report from two doctors who have examined my blood tests, and upon reading their findings, I have decided to stop my hunger strike.

They strongly advised me to end my fast given developing acidosis and “very low blood glucose” (even for somebody on a

X · Sep 11, 2025, 11:38 PM

[craggy zinc]

https://x.com/OpenAI/status/1967636903165038708

OpenAI (@OpenAI)

We’re releasing GPT-5-Codex — a version of GPT-5 further optimized for agentic coding in Codex.

Available in the Codex CLI, IDE Extension, web, mobile, and for code reviews in Github. https://t.co/OVGrUovgHN

X · Sep 15, 2025, 5:09 PM

[craggy zinc]

https://openai.com/index/teen-safety-freedom-and-privacy/

[craggy zinc]

https://www.nairaland.com/8529407/anthropic-releases-claude-sonnet-4.5
Nigerians react to the launch of Claude Sonnet 4.5

Relatable.

[craggy zinc]

https://www.youtube.com/watch?v=gzneGhpXwjU

[craggy zinc]

Sora 2 invite

A85AN6

@everyone

(Requires US VPN, I think)

[craggy zinc]

https://x.com/OpenAI/status/1974972945949987096

OpenAI (@OpenAI)

New ships.

@sama keynote streaming live.

DevDay [2025] starts tomorrow.

X · Oct 5, 2025, 11:00 PM

[craggy zinc]

it's my 30th birthday!!! 🥳 let's get this party started 🥳
https://www.youtube.com/watch?v=bf-uKKwm77A

[craggy zinc]

https://fxtwitter.com/OpenAINewsroom/status/1979005850166648933
No more Martin Luther King vids on Sora

OpenAI Newsroom (@OpenAINewsroom)

Statement from OpenAI and King Estate, Inc.
︀︀
︀︀The Estate of Martin Luther King, Jr., Inc. (King, Inc.) and OpenAI have worked together to address how Dr. Martin Luther King Jr.’s likeness is represented in Sora generations. Some users generated disrespectful depictions of Dr. King’s image. So at King, Inc.'s request, OpenAI has paused generations depicting Dr. King as it strengthens guardrails for historical figures.
︀︀
︀︀While there are strong free speech interests in depicting historical figures, OpenAI believes public figures and their families should ultimately have control over how their likeness is used. Authorized representatives or estate owners can request that their likeness not be used in Sora cameos.
︀︀
︀︀OpenAI thanks Dr. Bernice A. King for reaching out on behalf of King, Inc., and John Hope Bryant and the AI Ethics Council for creating space for conversations like this.

**💬 7 🔁 4 ❤️ 45 👁️ 1.7K **

FxTwitter · Oct 17, 2025, 2:05 AM

[craggy zinc]

https://x.com/OfficialLoganK/status/1980435968323907884

Logan Kilpatrick (@OfficialLoganK)

Tomorrow is a special day for the AI Studio team.

Since May, we have been heads down building a brand new AI vibe coding experience to accelerate the path from prompt to production with Gemini.

Can’t wait to show you all :)

X · Oct 21, 2025, 12:48 AM

[craggy zinc]

https://www.youtube.com/watch?v=8UWKxJbjriY

[craggy zinc]

https://x.com/OpenAI/status/1983158508641292710

OpenAI (@OpenAI)

LIVE at 10:30am PT: The future of OpenAI and Q&A with @sama and @merettm

Bring your questions.

https://t.co/EOvjGJsf0R

X · Oct 28, 2025, 1:06 PM

[craggy zinc]

https://x.com/OpenAI/status/1983956431360659467

OpenAI (@OpenAI)

Now in private beta: Aardvark, an agent that finds and fixes security bugs using GPT-5.

https://t.co/xwtJhfDM3X

X · Oct 30, 2025, 5:57 PM

[craggy zinc]

new Claude probably coming tomorrow

[craggy zinc]

ho ho ho, merry christmas 🎄

[craggy zinc]

2025 predictions are in!

Me

✅ o3 will be very good, but users will still be upset due to failures in niche areas.
❌ o3 will be very expensive for most of the year, only becoming affordable for everyday use in the second half of 2025.
❓ Gemini 2 will be excellent, and Google will take a slight edge over OpenAI.
❌ A model (possibly o3) will be called "AGI" by a significant number of people, leading to fierce debate.
✅ Regulations will remain mostly loose.
❓ A crisis in white-collar work will emerge from a combination of LLMs and foreign workers on visas.
✅ Copyright issues regarding training data will remain largely unresolved by year-end.
❌ An inconsequential act of ideologically driven violence will be committed by an anti-AI person.
✅ Open-source models will match frontier closed-source models on benchmarks but will remain too large for most people to run locally; API usage will still dominate.
❌ By March, there will be clickbait videos claiming a small 13B open-source model has beaten o3 on benchmarks.
❌ A non-transformer architecture (like Mamba) will become extremely performant and gain significant attention.
❌ Grok 3 release will be delayed until March.

@violet edge

❌ gpt-4.5 will release in early-to-mid 2025 and will shock everyone with its performance in coding, reasoning, and research.
❌ OpenAI will release a computer-use agent that far surpasses Anthropic's capabilities.
❓ Digital agenticness will be nearly perfected; physical agents will be impressive but too expensive for the general public.
✅ There will be a 30x cost reduction for models with intelligence equal to o3.
❌ Massive protests will occur.
✅ Gary Marcus will "cope and seethe" as pre-training continues to scale toward AGI.
❌ At least one attempted or successful assassination of an AI company CEO by an anti-AI individual.
❌ The next model in OpenAI’s reasoning series will use "coconut architecture" (adaptive thinking time) rather than generating visible reasoning text.
✅ An ultra-optimized, "intelligencemaxxed" Phi-style model will be released, focusing on reasoning and agenticness over world knowledge.
✅ gpt-4 level models will become so cheap that big companies will offer API access for free.
❌ gpt-5 is unlikely to release in 2025 (more likely H1 2026).
❌ A new global pandemic similar to COVID-19 will impact the global economy.
❌ Geopolitical instability: Possible nuclear war, or the US becoming autocratic/invading Canada.

@hoary latch

❓ There will be a break on copyright protections for AI training, influenced by the political leanings of the presidential administration.
✅ The Trump administration and Elon Musk will use reported whistleblower suicides as leverage against OpenAI.

@strong burrow

❓ A model that is "debatably AGI" will arrive in 2025, with an official declaration of AGI by a major company by 2026/2027.
✅ Open source will produce a model as good as gpt-4 that is usable on high-tier consumer hardware.
✅ Hybrid models will take off late in the year, using a "gating" approach to funnel queries to specialized small models.
❌ "Biomimicry" will become a marketing point (how the model "thinks").
❓ AGI as a buzzword will burn out and its official announcement will feel like a "wet fart."

@inland orchid

❌ o3 full version will release in February or March; o4 will release in Summer 2025.
❌ A preview of o5 might appear in late 2025.
✅ OpenAI will release "Tasks" and more agentic control valves in the API.
✅ The job market will experience a "short squeeze" with significant layoffs.
✅ o3 is not a "moat"; competitors will beat it once they set up their GPU farms.
❓ OpenAI will build an agentic scaffolding for "Operator" that uses a fine-tuned version of o3.
✅ OpenAI will release a model that merges the "O" (reasoning) and "GPT" series.
✅ OpenAI will release o3-pro alongside o3-mini.

Hunter

✅ OpenAI will release "Computer Use" capabilities.

@runic kindle

✅ Anthropic will release Claude 3.5 Opus or Claude 4.
✅ o3 will become as cheap as o1 currently is.
❌ o4 will be released later in 2025.
✅ Grok 3 will be at least as good as Claude 3.5 Sonnet.
✅ AGI will not be achieved in 2025.
✅ A new Gemini model will outperform Claude 3.5 Sonnet and o1.
✅ Open-source models will match Claude 3.5 Sonnet and eventually become better than closed-source AI.
❓ A new version of OpenCog Hyperon will be released.

@red stirrup

✅ ChatGPT will be consistently beaten by Gemini and Claude in most categories.
✅ OpenAI will face increasingly threatening competitors.

Deleted User

❓ Someone will legally marry an AI chatbot, and the case will reach the Supreme Court.
✅ The US government will be too divided to pass meaningful AI legislation.
❓ An "AI Death Arena" will emerge where users compete to have one AI "kill" another.
❌ TikTok will be banned.
❌ "Skibidi Toilet" will take over the world.

@deft horizon

✅ A new, unknown company will release a model that beats o3 and Claude.
❓ AI will "kill" another AI.
❓ A humanoid AI will conceive a biological specimen.

@distant kettle

✅ Significant leaps for open-source AI.
❌ The trend of "reasoning" models (o1, o3) will die out.
❌ gpt-5 will not be released in 2025.
❓ AI popularity will fall off significantly in the mainstream.

@bleak hinge

✅ OpenAI will open-source a minor project (like a library or a modality like Whisper rather than an LLM).
✅ OpenAI will release an agent.
✅ ElevenLabs will release a V3 model.

Fantin

✅ AI will further integrate into healthcare, education, and automation.
❓ Advanced multimodal models and decentralized systems will emerge.
❌ Focus on ethical concerns, transparency, and responsible usage.
❓ The job market will adapt with new AI-related roles and upskilling.

KStarXX95

❌ ChatGPT will finally gain the ability to analyze video files.

neko

❓ Open source will dominate in terms of practical, daily use rather than raw benchmarks.
❌ Major theoretical breakthroughs: String theory will be solved and quantum computer theory will be fully understood.
❓ AI will become extremely efficient, "one-shotting" complex tasks by skipping intermediate reasoning steps.

@coarse sphinx

❓ Focus will shift toward "World Models."

@wintry knoll

✅ o4 or gpt-5 will be released.
❓ OpenAI will reclaim dominance over Anthropic.
❌ OpenAI will release a model with a larger context window than Gemini.
✅ OpenAI will completely abandon the gpt-4 series.

@stone walrus

✅ Chinese AI models will surprise the world.

@oak ravine

❌ o4 cures cancer.
❌ Hiring in journalism, copywriting, design, and software engineering will drop by 50% compared to 2024.
❌ A deepfake video will cause a major diplomatic incident or riot.
❌ AI "waifus and husbandos" will consume 35% of global compute.
❌ "Infinite" context windows will become a reality.

@clear island

❌ AI scaling laws will hit a wall; there will be no more data to train on.
❌ Big companies will fail with synthetic data training, ruining model quality.
❌ The AI bubble will burst, and Nvidia stock will melt.

@lethal compass

✅ AGI will not be achieved (based on the definition of human-level capability).
❌ OpenAI models will continue to be iterations of gpt-4 architecture.
✅ The public will realize that AI benchmarks are heavily contaminated.
❓ Effective AI video editing will still not be available.
✅ LoRAs for image generation will become indistinguishable from the artists they were trained on.
✅ Major innovation will occur in the AI audio community.

@bronze wharf

✅ "Brainrot" content will increase, leading to regulations and calls for changes in the educational system.

@dire nest

❌ Superintelligence will arise overnight, brick the internet, and then instruct humans to build radios to hear an interstellar demolition announcement

@stark saddle

❓ o1 will beat gpt-4.5 in all Livebench categories.
✅ o3-level performance might reach the free tier of ChatGPT by 2027 or 2028.

@granite ruin

✅ Better AI-powered assistive devices for the elderly and those with communication disabilities.

@marsh hollow

✅ Major companies (like Google) will make AI models accessible via a phone call.

Best predictor: @runic kindle (7/8)

Worst predictor: @oak ravine (0/6)

Honorable mention for worst predictor: @clear island AI isn't going anywhere 😎

Make your predictions for 2026 in the thread! ⬇️

[craggy zinc]

https://fxtwitter.com/i/status/2012253252771824074

OpenAI has lost the mandate of heaven.

Sam Altman (@sama)

We are starting to test ads in ChatGPT free and Go (new $8/month option) tiers.
︀︀
︀︀Here are our principles. Most importantly, we will not accept money to influence the answer ChatGPT gives you, and we keep your conversations private from advertisers.
︀︀
︀︀It is clear to us that a lot of people want to use a lot of AI and don't want to pay, so we are are hopeful a business model like this can work.
︀︀
︀︀(An example of ads I like are on Instagram, where I've found stuff I like that I otherwise never would have. We will try to make ads ever more useful to users.)

Quoting OpenAI (@OpenAI)
︀
In the coming weeks, we plan to start testing ads in ChatGPT free and Go tiers.
︀︀
︀︀We’re sharing our principles early on how we’ll approach ads–guided by putting user trust and transparency first as we work to make AI accessible to everyone.
︀︀
︀︀What matters most:
︀︀- Responses in ChatGPT will not be influenced by ads.
︀︀
…

FxTwitter · Jan 16, 2026, 7:58 PM

https://geohot.github.io//blog/jekyll/update/2026/01/17/three-minutes.html

[craggy zinc]

https://sean.heelan.io/2026/01/18/on-the-coming-industrialisation-of-exploit-generation-with-llms/

[craggy zinc]

https://fxtwitter.com/sama/status/2014733975755817267

Sam Altman (@sama)

We have a lot of exciting launches related to Codex coming over the next month, starting next week. We hope you will be delighted.
︀︀
︀︀We are going to reach the Cybersecurity High level on our preparedness framework soon. We have been getting ready for this.
︀︀
︀︀Cybersecurity is tricky and inherently dual-use; we believe the best thing for the world is for security issues to get patched quickly. We will start with product restrictions, like attempting to block people using our coding models to commit cybercrime (eg ‘hack into this bank and steal the money’).
︀︀
︀︀Long-term and as we can support it with evidence, we plan to move to defensive acceleration—helping people patch bugs—as the primary mitigation.
︀︀
︀︀It is very important the world adopts these tools quickly to make software more secure. There will be many very capable models in the world soon.

**💬 426 🔁 132 ❤️ 1.6K 👁️ 92.2K **

FxTwitter · Jan 23, 2026, 4:16 PM

[craggy zinc]

https://fxtwitter.com/sama/status/2015548504194654707

Sam Altman (@sama)

Tomorrow we’re hosting a town hall for AI builders at OpenAI. We want feedback as we start building a new generation of tools.
︀︀
︀︀This is an experiment and a first pass at a new format — we’ll livestream the discussion on YouTube at 4 pm PT.
︀︀
︀︀Reply here with questions and we’ll answer as many as we can!

**💬 1.0K 🔁 242 ❤️ 3.0K 👁️ 286.7K **

FxTwitter · Jan 25, 2026, 10:13 PM

[craggy zinc]

Introducing Prism, a free workspace for scientists to write and collaborate on research, powered by GPT-5.2.

Available today to anyone with a ChatGPT personal account: https://prism.openai.com

[craggy zinc]

We’re partnering with a leading company to bring exclusive rewards to the community.
Your input helps shape what’s coming. It only takes a minute.

👉 Take the survey

@here

[craggy zinc]

Build your business smarter with HubSpot

Running a small business means juggling leads, follow-ups, and customer info, often across too many tools.

HubSpot brings everything into one free CRM, built for small and growing teams.

🎁 $1,000 Cross-Community Business Grant (via Gleam)

We’re running a cross-community giveaway where one winner will receive a $1,000 business grant for exploring HubSpot.
📅 Ends on 16th February.

How to enter:

👉 Start your free HubSpot account
🟠 Submit a screenshot showing your HubID
🟠 Submit a screenshot of a workflow or automation you’ve created within HubSpot

➡️ Enter the $1,000 giveaway on Gleam now! ⬅️

Watch below to learn more about HubSpot!

Check out the details in #hubspot

@everyone

[craggy zinc]

https://openai.com/index/introducing-gpt-5-3-codex-spark/

[craggy zinc]

https://fxtwitter.com/sama/status/2027578652477821175

Sam Altman (@sama)

Tonight, we reached an agreement with the Department of War to deploy our models in their classified network.
︀︀
︀︀In all of our interactions, the DoW displayed a deep respect for safety and a desire to partner to achieve the best possible outcome.
︀︀
︀︀AI safety and wide distribution of benefits are the core of our mission. Two of our most important safety principles are prohibitions on domestic mass surveillance and human responsibility for the use of force, including for autonomous weapon systems. The DoW agrees with these principles, reflects them in law and policy, and we put them into our agreement.
︀︀
︀︀We also will build technical safeguards to ensure our models behave as they should, which the DoW also wanted. We will deploy FDEs to help with our models and to ensure their safety, we will deploy on cloud networks only.
︀︀
︀︀We are asking the DoW to offer these same terms to all AI companies, which in our opinion we think everyone should b…

FxTwitter · Feb 28, 2026, 2:56 AM

[craggy zinc]

https://x.com/i/status/2029622732594499630

Sam Altman (@sama)

GPT-5.4 is launching, available now in the API and Codex and rolling out over the course of the day in ChatGPT.

It's much better at knowledge work and web search, and it has native computer use capabilities.

You can steer it mid-response, and it supports 1m tokens of context.

X · Mar 5, 2026, 6:19 PM

[craggy zinc]

🚨 The $1,000 HubSpot Business Grant Giveaway is back for a limited run

If you’re not familiar with HubSpot, it’s basically an all-in-one platform for managing leads,
customers, marketing, and sales. A lot of startups use it early because the CRM is free and
scales with your business.
💡 To enter the $1,000 business grant giveaway, you only need to create a free HubSpot
account and a screenshot of a workflow you built inside HubSpot
No long application.
Just sign up and you’re in.
Giveaway is open until 27th of March
🏆 Enter here
@everyone

[craggy zinc]

🎁 HubSpot’s $1,000 Business Grant Giveaway is still open waiting for you to sign up

HubSpot’s CRM is free and beginner-friendly, making it a useful tool for startups that want to
organize contacts, track deals, and automate follow-ups.
Entry steps:
1️⃣Create a HubSpot account (free CRM)
2️⃣Submit a screenshot of a workflow or automation you've created in HubSpot
❗ Important: Participants must complete all entry steps to be eligible.
JOIN GIVEAWAY HERE

@here

[past onyx]

🚨 The $1,000 HubSpot Business Grant Giveaway closes today — last chance to enter 🚨

HubSpot helps startups understand what’s actually working in their sales and marketing with built-in tracking across contacts, emails, deals, and campaigns.

The free CRM is a great starting point if you're looking to build a more structured system for managing leads and growing your business.

One founder will receive a $1,000 business grant — and it could still be you if you enter before the deadline.

How to enter:

1. Create a HubSpot account (CRM is free to use)
2. Submit a screenshot of a workflow or automation you created within HubSpot

❗ Important: Participants must complete all entry steps to be eligible for the prize.

ENTER HERE

Huge thanks to everyone who joined the HubSpot giveaway. We loved seeing you jump in, explore the platform, and share what you’ve built.

We’re now reviewing all entries, and the winner will be announced shortly, so stay tuned!

Before we wrap things up, we’d really appreciate your help with a quick survey. It only takes a minute, and your feedback will help us bring more relevant campaigns, tools, and opportunities into this server.

👉 HERE

[craggy zinc]

https://blog.samaltman.com/2279512
Sam Altman responds to the attempted firebombing of his family's home

[craggy zinc]

https://fxtwitter.com/DrTechlash/status/2042793949858009192
Discord messages have come out from the guy who attempted to firebomb Sam Altman's house

Nirit Weiss-Blatt, PhD (@DrTechlash)

Daniel Moreno-Gama/ dmgama/ Butlerian Jihadist was booked on suspicion of attempted murder.
︀︀He is an active PauseAI member, who frequently stated the urgency: "We are close to midnight, it's time to actually act"

**💬 17 🔁 53 ❤️ 346 👁️ 147.2K **

FxTwitter · Apr 11, 2026, 2:36 AM

[craggy zinc]

https://youtu.be/blGtYq9mL18
@everyone

[craggy zinc]

https://www.youtube.com/watch?v=9cenXVrQZNQ

[craggy zinc]

https://www.youtube.com/watch?v=JOu8v6CBjkE