Something is restricting me to a much smaller context | OpenRouter | Page 1

teal mulch Sep 21, 2025, 9:42 PM

#

The is the error message I keep getting is:
{"message":"Prompt tokens limit exceeded: 13880 > 2021. To increase, visit https://openrouter.ai/settings/credits and upgrade to a paid account","code":402,"metadata":{"provider_name":null}}

And if I try to use Claude I get this:
PROXY ERROR 402: {"error":{"message":"This request requires more credits, or fewer max_tokens. You requested up to 500 tokens, but can only afford 168. To increase, visit https://openrouter.ai/settings/credits and upgrade to a paid account","code":402,"metadata":{"provider_name":null}}

I am getting a limit of ~2,021 tokens for prompts when I use Gemini and ~4,593 tokens for prompts when I use minimax.
2,021 is much less than the 1,048,576 possible for the Gemini-2.5-Pro model that I was just trying to use. The amount I got for minimax is also much less than the possible amount for it. And I am not even sure what is going on with Claude.

Something is restricting me to a much smaller context in my situation and I **really **want to get rid of that something.

Here is some more context from my situation:

my current credit balance is 9,94 so I already have a paid account, if I have understood correctly
I use the models for chatbots from janitor.ai
janitor.ai does not really have a context limit even though after 16384 it starts to forget stuff
after I got the error message for the first time I used a different model for a while before trying to go back to the Gemini, Claude or minimax again
the models I have had this problem with are google/gemini-2.5-pro, anthropic/claude-3-7-sonnet, minimax/minimax-m1
for now, I have not had this problem with mistralai/mistral-nemo
I may have this problem with even more models but these are the ones I have tested
my typical usage is about ~100 tokens input and ~390 tokens output

Also, I do not understand all the difficult terminology yet so please use simpleish language :3

placid irisBOT Sep 21, 2025, 9:42 PM

#

I'm an AI assistant - my answers may be incorrect and I cannot see screenshots. Please verify information with the OpenRouter docs.

You are seeing lower token/context limits because of your current credit balance and/or pricing tier. The error message says:

"Prompt tokens limit exceeded: 13880 > 2021. To increase, visit https://openrouter.ai/settings/credits and upgrade to a paid account"

This means your account is being limited to a much smaller number of tokens per request because of your available credits or account level. Even if you have some credits, the system may limit your max context size based on your balance and model pricing. For bigger context windows (like the full 1,048,576 tokens possible for google/gemini-2.5-pro), you must have enough credits to cover the cost of those tokens in a single request.

For example, the Gemini 2.5 Pro model supports up to 1.05M tokens context, but you will be limited by your credits and possibly other account limits.
Claude 3.7 Sonnet supports 200,000 context tokens, but again, your actual request limit depends on your available credits and usage.

If you want to use larger context windows, you need a higher credit balance. If you keep running into these limits despite having credits, your available credits may not be enough to cover the cost of the large context window you are trying to use. You can check your credits and upgrade your account at OpenRouter Credits Settings Google Vertex | OpenRouter.

If you believe this is an error and you have enough credits, or if you want to discuss...

Author Not Found

The author you are looking for could not be found.

Gemini 2.5 Pro Experimental - API, Providers, Stats

This model has been deprecated by Google in favor of the (paid Preview model)[google/gemini-2.5-pro-preview]

Gemini 2. Run Gemini 2.5 Pro Experimental with API

Claude 3.7 Sonnet - API, Providers, Stats

Claude 3.7 Sonnet is an advanced large language model with improved reasoning, coding, and problem-solving capabilities. Run Claude 3.7 Sonnet with API

Provider Not Found | OpenRouter

The provider you are looking for could not be found.

#

... account-specific issues, please email [email protected]. We do not handle billing or account issues on Discord.

Let me know if you have more questions or need help understanding the pricing or context limits for a specific model!

-# This is an automated response powered by Answer HQ. Use !mute to stop the bot from responding.

teal mulch Sep 21, 2025, 9:53 PM

#

So how much higher credit balance I need to buy in order to get the models full possible amount, if 10 credits is not enough.
Or at least how much more I need to buy for a 16384 context window?

blissful peak Sep 21, 2025, 10:23 PM

#

teal mulch So how much higher credit balance I need to buy in order to get the models full ...

models are cheap these days! even the minimum payment should get you thousands-millions of tokens depending on the model

teal mulch Sep 21, 2025, 10:30 PM

#

blissful peak models are cheap these days! even the minimum payment should get you thousands-m...

Yeah true. But Something is restricting me to a much smaller context window. I can still use tokens but only for few messages before the total prompt being sent to the model gets too large. The issue is that I am not fully sure about what is restricting me to a much smaller context.

blissful peak Sep 21, 2025, 10:30 PM

#

anything of note on https://openrouter.ai/settings/credits or https://openrouter.ai/activity?

teal mulch Sep 21, 2025, 10:35 PM

#

blissful peak anything of note on <https://openrouter.ai/settings/credits> or <https://openrou...

At least not to my knowledge. Everything seems normal after checking.

blissful peak Sep 21, 2025, 10:35 PM

#

what's your balance at?

#

the 9.94 from earlier?

teal mulch Sep 21, 2025, 10:35 PM

#

9,94

#

ye

blissful peak Sep 21, 2025, 10:36 PM

#

oh that's really weird

teal mulch Sep 21, 2025, 10:36 PM

#

ikr

blissful peak Sep 21, 2025, 10:36 PM

#

does the same problem happen in the openrouter chatroom?

teal mulch Sep 21, 2025, 10:38 PM

#

I doubt it would not but I'll maybe check later. I'll have to go asleep now. It is way after midnight on my time zone.

teal mulch Sep 22, 2025, 7:46 AM

#

Does anyone know a solution to my problem? I still have not been able to figure it out

teal mulch Sep 22, 2025, 9:22 AM

#

<@&1384697330254610442> Could someone help me at some point? This is not overly urgent but this has been going on for few days already for me.

flat dirge Sep 22, 2025, 11:25 PM

#

teal mulch <@&1384697330254610442> Could someone help me at some point? This is not overly ...

can you post the new error in full?

teal mulch Sep 23, 2025, 6:13 AM

#

flat dirge can you post the new error in full?

This is the error code I am getting with most models: PROXY ERROR 402: {"error":{"message":"Prompt tokens limit exceeded: 9279 > 790. To increase, visit https://openrouter.ai/settings/credits and upgrade to a paid account","code":402,"metadata":{"provider_name":null}},"user_id":"user_32pvTI8UgNuynt7YlGQ2PDacGN9"} (unk)

Or then the models work, but only respond with a few words. But DeepSeek V3.1 has worked even though most models have not.

hard gust Sep 23, 2025, 8:22 AM

#

teal mulch This is the error code I am getting with most models: PROXY ERROR 402: {"error":...

Does this happen in the OR chat? https://openrouter.ai/chat

Can you post a screenshot of your balance https://openrouter.ai/settings/credits

OpenRouter

A router for LLMs and other AI models

teal mulch Sep 23, 2025, 10:09 AM

#

hard gust Does this happen in the OR chat? https://openrouter.ai/chat Can you post a scre...

This is my balance. And I am not sure if my context window would be this small in the OR chat too because I use the models only on janitor ai by copypasting the needed information in the api settings like this.

Would it be useful if I test my luck in the open router chats with some model that I have had this problem with to see if the prompt tokens limit is smaller than it should be in there as well.

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..png

flat dirge Sep 23, 2025, 10:55 AM

#

teal mulch This is my balance. And I am not sure if my context window would be this small i...

Yes, it would be helpful

winter igloo Sep 23, 2025, 10:56 AM

#

Did you set a credit limit on your api key? you can check it on this page https://openrouter.ai/settings/keys

teal mulch Sep 23, 2025, 11:03 AM

#

winter igloo Did you set a credit limit on your api key? you can check it on this page https:...

no I have not.

teal mulch Sep 23, 2025, 11:03 AM

#

flat dirge Yes, it would be helpful

I'll check

teal mulch Sep 23, 2025, 11:27 AM

#

I used DeepSeek: DeepSeek V3 0324 on OR chat and send it a text about photosynthesis and told it to summarize it and also tell me how much prompt tokens I have used. So it works in open router chat.

flat dirge Sep 23, 2025, 11:29 AM

#

teal mulch I used DeepSeek: DeepSeek V3 0324 on OR chat and send it a text about photosyn...

DeepSeek doesn't know how many tokens were used and isn't self aware, but it's weird that it works in the chat room, but not in JAI

flat dirge Sep 23, 2025, 11:30 AM

#

teal mulch This is the error code I am getting with most models: PROXY ERROR 402: {"error":...

Try creating a new API key

teal mulch Sep 23, 2025, 12:39 PM

#

flat dirge DeepSeek doesn't know how many tokens were used and isn't self aware, but it's w...

Yeah I know, It generates responses by analyzing probabilities in language, not by reasoning or awareness. I told it to estimate the amount of tokens used in the text I sended.

teal mulch Sep 23, 2025, 12:40 PM

#

flat dirge Try creating a new API key

I have tried that few times already. I'll try making a new account and see if it works then.

#

But this is not supposed to happen, yes? Should I be able to have larger prompt token amount in jai

flat dirge Sep 23, 2025, 1:01 PM

#

teal mulch But this is not supposed to happen, yes? Should I be able to have larger prompt ...

Yea

#

I honestly have no clue

#

that is so weird

#

cc @true wharf

#

TLDR;

User can't send prompts larger than a few hundred tokens
Error message from JAI
PROXY ERROR 402: {"error":{"message":"Prompt tokens limit exceeded: 9279 > 790. To increase, visit https://openrouter.ai/settings/credits and upgrade to a paid account","code":402,"metadata":{"provider_name":null}},"user_id":"user_32pvTI8UgNuynt7YlGQ2PDacGN9"} (unk)

There is no error in the OpenRouter chatroom

The API key has been deleted and recreated multiple times
The user has enough money (more than $9)

true wharf Sep 23, 2025, 1:14 PM

#

that user ID does not have money in it

#

the API key you're creating is on the wrong account or something

#

#

flat dirge Sep 23, 2025, 1:46 PM

#

@teal mulch

teal mulch Sep 23, 2025, 2:25 PM

#

@flat dirge In the past I have made two accounts from two differend emails but only putted money in one of them and that is the one I have been using, the keys are from that account

#

They at least most certainly should be from the account that has money in it. I'll double check when I get home in about 45 minutes

teal mulch Sep 23, 2025, 3:34 PM

#

Well, that was it.
When I have made new keys I have always copied them on a note app on my phone first but it seems I had not deleted the old ones from there so somehow I had accidentally ended up using the old keys instead. I am sorry for the inconvenience I caused.
Thank you both for your help :).

#Something is restricting me to a much smaller context