Skip to main content
Code: max_input_tokens_exceeded HTTP status: 400 Type: invalid_request

​
When it fires

The total tokens on your chat or embedding request (system + prior turns + current input, including multimodal parts) exceed the model’s aurous_metadata.context_window. No charge — any held credits are released.

​
How to recover

Trim the input (shorter system prompt, fewer prior turns, smaller images / videos, lower detail), or switch to a model with a larger context window. The window is on the model row at GET /v1/models.