[Bug]: Unable to use Claude Opus 4.7 in Claude Code

[saraangelmurphy]

Check for existing issues

• I have searched the existing issues and checked that my issue is not a duplicate.

What happened?

A bug happened!

Using Claude Opus 4.7 in claude code through liteLLM consistently fails with the error:

❯ hello world
  ⎿  API Error: 400 {"error":{"message":"b'{\"type\":\"error\",\"error\":{\"type\":\"invalid_request_error\",\"message\":\"\\\\\"thinking.type.enabled\\\\\" is not supported for this model. Use \\\\\"thinking.type.adaptive\\\\\" and \\\\\"output_config.effort\\\\\" to
     control thinking behavior.\"},\"request_id\":\"req_011Ca7qUXxE8LhWBqa2KebcM\"}'","type":"None","param":"None","code":"400"}}

The model works fine when curling it directly. The same error occurs even in anthropic's passthrough mode (i.e. when I add /anthropic to my ANTHROPIC_BASE_URL.

➜  scratch curl -s --location https://<redacted>/chat/completions --header 'Content-Type: application/json' --header "Authorization: Bearer $LITELLM_PROD_API_KEY" --data @data.json  | jq .
{
  "id": "chatcmpl-83449e15-cd2a-483d-ba51-d9a7593b7e3a",
  "created": 1776360820,
  "model": "claude-opus-4-7",
  "object": "chat.completion",
  "choices": [
    {
      "finish_reason": "stop",
      "index": 0,
      "message": {
        "content": "Pi to four decimal places is **3.1416**.",
        "role": "assistant",
        "provider_specific_fields": {
          "citations": null,
          "thinking_blocks": null
        }
      }
    }
  ],
  "usage": {
    "completion_tokens": 21,
    "prompt_tokens": 21,
    "total_tokens": 42,
    "completion_tokens_details": {
      "reasoning_tokens": 0,
      "text_tokens": 21
    },
    "prompt_tokens_details": {
      "cached_tokens": 0,
      "cache_creation_tokens": 0,
      "cache_creation_token_details": {
        "ephemeral_5m_input_tokens": 0,
        "ephemeral_1h_input_tokens": 0
      }
    },
    "cache_creation_input_tokens": 0,
    "cache_read_input_tokens": 0
  }
}

Steps to Reproduce

1. Add Claude Opus 4.7 to the model cost map. I used both Anthropic and Vertex and the issue occurs with both.

- litellm_params:
    api_key: os.environ/ANTHROPIC_API_KEY
    model: anthropic/claude-opus-4-7
  model_info:
    mode: chat
  model_name: claude-opus-4-7
- litellm_params:
    cache_control_injection_points:
    - location: message
      role: system
    model: vertex_ai/claude-opus-4-7
    use_in_pass_through: true
    vertex_credentials: os.environ/GOOGLE_SA_KEYFILE_PATH
    vertex_location: global
    vertex_project: <redacted>

2. Open Claude Code (v2.1.92)
3. Set the model to claude-opus-4-7 with /model claude-opus-4-7
4. Submit any message and receive the error mentioned above.

  ⎿  API Error: 400 {"error":{"message":"b'{\"type\":\"error\",\"error\":{\"type\":\"invalid_request_error\",\"message\":\"\\\\\"thinking.type.enabled\\\\\" is not supported for this model. Use \\\\\"thinking.type.adaptive\\\\\" and \\\\\"output_config.effort\\\\\" to
     control thinking behavior.\"},\"request_id\":\"req_011Ca7qUXxE8LhWBqa2KebcM\"}'","type":"None","param":"None","code":"400"}}

5. The same issue does not occur when using Claude Code without going through LiteLLM, or when using Opus 4.6 (or any other anthropic model)

Relevant log output

fastapi.exceptions.HTTPException: 400: b'{"type":"error","error":{"type":"invalid_request_error","message":"\\"thinking.type.enabled\\" is not supported for this model. Use \\"thinking.type.adaptive\\" and \\"output_config.effort\\" to control thinking behavior."},"request_id":"req_011Ca7qUXxE8LhWBqa2KebcM"}'

What part of LiteLLM is this about?

Proxy

What LiteLLM version are you on ?

v1.82.3

Twitter / LinkedIn details

No response

[mbegur]

Upgrading to the latest version of claude code fixes this for me

[wenshiqi0]

@mbegur which version may I ask

[mbegur]

v2.1.111

[saraangelmurphy]

this is indeed resolved by using v2.1.111, which is now available in homebrew using claude-code@latest (not in claude-code which is the stable version and is behind by about a week) .

[andrii29]

The problem persists on Linux with AWS Bedrock.
Claude CLI v2.1.111 and v2.1.112
LiteLLM version v1.82.3-stable.patch.4

API Error: 400 {"error":{"message":"{\"message\":\"{\\\"type\\\":\\\"error\\\",\\\"request_id\\\":\\\"req_redacted\\\",\\\"error\\\":{\\\"type\\\":\\\"invalid_request_error\\\",\\\"message\\\":\\\"\\
     \\\\\"thinking.type.enabled\\\\\\\" is not supported for this model. Use \\\\\\\"thinking.type.adaptive\\\\\\\" and \\\\\\\"output_config.effort\\\\\\\" to control thinking behavior.\\\"}}\"}. Received Model                                    
     Group=example-bedrock-claude-4-7-opus\nAvailable Model Group Fallbacks=None","type":"None","param":"None","code":"400"}}

litellm config

model_list:
  - model_name: example-bedrock-claude-4-7-opus
    litellm_params:
      model: 'bedrock/eu.anthropic.claude-opus-4-7'
      aws_access_key_id: os.environ/AWS_ACCESS_KEY_ID
      aws_secret_access_key: os.environ/AWS_SECRET_ACCESS_KEY
      aws_region_name: os.environ/AWS_REGION_NAME