I can not configure the limit for litellm

[m-da-costa]

Sorry if my question is a basic one but I cannot configure the context window (limits) for my litellm instance. Se examples:
config.json

    "litellm": {
      "npm": "@ai-sdk/openai-compatible",
      "options": {
        "baseURL": "https://my.domain.com/v1"
      },
      "models": {
        "cerebras/qwen-3-coder-480b": {
          "tools": true,
          "reasoning": true,
          "context_window": 120000,
          "limit": {
            "context": 120000,
            "output": 64000
          }
        },
      }
    }

[github-actions]

This issue might be a duplicate of existing issues. Please check:

• AI_APICallError with Qwen 3 Coder 480b: Please reduce the length of the messages or completion. Current length is 69553 while limit is 65536 #1551: Similar issue with Qwen 3 Coder 480b where context limits are not properly respected, causing "Please reduce the length of the messages or completion" errors
• max_tokens defaults to 32000 when using a custom provider #1735: Reports that max_tokens defaults to 32000 when using custom providers with openai-compatible, and discusses how limit configuration in config.json should work as a workaround
• Model options are not forwarded for openai-compatible providers unless provider.name option is given #971: Model options are not forwarded for openai-compatible providers unless provider.name option is given, which may affect limit configurations

Feel free to ignore if none of these address your specific case.

[rekram1-node]

@m-da-costa put:

includeUsage: true

in the provider options

ex:

 "litellm": {
      "npm": "@ai-sdk/openai-compatible",
      "options": {
        "baseURL": "https://my.domain.com/v1",
        "includeUsage": true
      },
      "models": {
        "cerebras/qwen-3-coder-480b": {
          "tools": true,
          "reasoning": true,
          "context_window": 120000,
          "limit": {
            "context": 120000,
            "output": 64000
          }
        },
      }
    }

[rekram1-node]

going forward we will set this automatically but until next release you should be able to follow waht I mentioned above^^

[m-da-costa]

Ty so much @rekram1-node