[✅ Solved] Responses from ai on Janitor AI is very short

[eternal canopy]

Thanks for sharing your settings and detailed description! The issue of short responses from the AI is most often related to the chat history and prompt configuration, not the API itself, especially since you haven’t changed temperature or other limits.

Here are some things to check and try:

1. Remove Jailbreak/Custom Prompts:
If you have ever used a jailbreak, special instructions, or custom prompt in your chats, these remain in the conversation history and can cause the AI to become brief or strange in its replies.
Solution:

• Start a completely new chat with the character (not just refresh).
• Make sure the prompt/jailbreak field contains only a single space.

2. Long Previous Messages:
The model’s output length is influenced by your conversation history. If your past replies/messages are short, the AI will often start producing shorter responses too.
Solution:

• Try writing a longer message, or reset the conversation!

3. JanitorAI "Max Tokens" Setting:
Your "max tokens" is set to 0, which should be unlimited, but sometimes the site doesn’t handle this as expected and may cap your length.
Solution:

• Set "max tokens" to a large number (like 1000) just in case.

4. Website Issue:
Occasionally, JanitorAI itself may have issues unrelated to the API. If none of the above fix your problem, it could be a temporary issue with JanitorAI.

---

If you still have issues after trying the above:

• Try another website (like VenusChub or VenusChat.AI) to see if the problem persists.
• If you ever used a jailbreak or custom prompt, always reset the chat.

Reminder: Our models are unfiltered, so you don’t need jailbreaks or extra instructions.

Check the Models Channel for more tips and recommended settings. If the problem remains, wait for community helpers or create a ticket for further support.

Please note: This is an automated response from @eternal canopy. While it strive to provide accurate assistance, It may occasionally make mistakes. If you find any inaccuracies or need further clarification, feel free to wait for our community helpers to give you further guidance. Thank you.

[weak salmon]

Changing the context size wouldn't change anything in the bots responses it just sets the size on how many tokens the bot can remember before it will start to forget past messages.

Also set your context size to about 16384 otherwise you'll run into a context size issue.

What i can suggest is to edit the bot messages into longer paragraphs yourself until it learns and to write in long paragraphs aswell, also trying it out on different sites gives different responses aswell either better or worse due to how the websites are programmed for their own services which makes using the cosmos model on it as a third party option into giving out worse responses due to the sites programming.

And theres also the factor of how the bot was prompted by the creator themselfs, high token count doesn't always mean its gonna be a good bot most likely just a lot of bloat and perhaps repetitive prompts are inside such bots personality fields.

Other than that i don't think there is anything i can suggest or how to explain the situation many users have with the model

P.S: anyones free to correct me of i said smth wrong or that needs to be written better

[misty raptor]

Even on bots I was already using before it started doing it which is why I am very confused. I’ll just try later, but I will fix the context size. Thank you for responding

[sonic marsh]

#📢𝐀𝐧𝐧𝐨𝐮𝐧𝐜𝐞𝐦𝐞𝐧𝐭 message should give you your answer

[indigo zinc]

On janitor specifically, try setting max tokens to 1000

In my experience on that site 0 is not actually unlimited

[eternal canopy]

[✅ Solved] Responses from ai on Janitor AI is very short