OpenAI, Anthropic, Google, and most OpenAI-compatible providers

"Context length exceeded" / "prompt is too long" — fixing oversized AI requests

{"error":{"message":"This model's maximum context length is 200000 tokens. However, your messages resulted in 214880 tokens.","type":"invalid_request_error","code":"context_length_exceeded"}}

What does "context length exceeded" mean?

The prompt plus the requested output is larger than the model's context window. The model never starts generating — the request is rejected up front.

Why does it happen?

Long agent conversations, large pasted files, or a max_tokens value that leaves no room once the prompt is counted. Clients that guess a model's window from its name frequently guess high and overshoot.

How do I fix it?

1. 1

   Compact or restart the conversation so the history stops growing.

2. 2

   Lower max_tokens — the output reservation counts against the same window.

3. 3

   Switch to a larger-context model for that turn; kRouter can route a single request elsewhere without changing your setup.

4. 4

   Let your client read the real limits: kRouter publishes context_length and max_completion_tokens on /v1/models, so tools stop guessing from the model name.

Is it really this error?

These get mistaken for each other, and chasing the wrong one is where the time goes.

What will not fix it

Reasonable-sounding fixes that do not apply here — worth ruling out before you spend an afternoon on them.

• Retrying unchanged.

  Token counting is deterministic. The same messages produce the same overflow every time.

• Raising max_tokens to make room.

  Backwards — the output reservation is counted inside the same window, so raising it leaves less space for your prompt.

What kRouter does automatically

It does NOT try your other accounts, because the same request would fail identically on all of them — the error is returned to you straight away instead of burning every credential you have.