Repository navigation
Replies: 1 comment 2 replies
|
This error occurs when 9Router forwards your request to the upstream AI provider (e.g. Anthropic, Gemini, or OpenAI), and the upstream model API rejects the payload with HTTP 400. In long-running CLI sessions where work "gets stuck and then appears again", this typically stems from two causes: 1. Context Window / Token Limit OverflowAs a CLI session progresses, accumulated message history, tool execution results, and file diffs inflate the prompt payload. Once the token count exceeds the upstream model's context window limit (or maximum output limit), the upstream provider rejects the call with Fix:
2. Thinking / Temperature Parameter Conflict in v0.5.119Recent updates forwarded extended thinking configurations ( Troubleshooting Steps:
|
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
getting this error after update to 0.5.119
when using cli some work will get stuck and get this error, than need to start new session and appear again
API Error: 400 [400]: {"error":{"message":"The upstream model rejected the request. Check the request parameters.","type":"invalid_request_error","code":"upstream_rejected_request"}}
All reactions