cc 使用 zai-glm-4.7 报错 body.reasoning

[imchao9]

[debug] [openai_compat_executor.go:240] request error, error status: 422, error body: {"message":"body.reasoning_effort: Input should be 'low', 'medium' or 'high'\nbody.reasoning: property 'body.reasoning' is unsupported","type":"invalid_request_error","param":"validation_error","code":"wrong_api_format"}

配置如下：

# Optional payload configuration
payload:
  # default: # Default rules only set parameters when they are missing in the payload.
    # - models:
    #     - name: "gemini-2.5-pro" # Supports wildcards (e.g., "gemini-*")
    #       protocol: "gemini" # restricts the rule to a specific protocol, options: openai, gemini, claude, codex
    #   params: # JSON path (gjson/sjson syntax) -> value
    #     "generationConfig.thinkingConfig.thinkingBudget": 32768
  override: # Override rules always set parameters, overwriting any existing values.
    - models:
        - name: "zai-glm-4.7" # Supports wildcards (e.g., "gpt-*")
          protocol: "openai" # restricts the rule to a specific protocol, options: openai, gemini, claude, codex
      params: # JSON path (gjson/sjson syntax) -> value
        "reasoning.effort": "high"

大佬们这个如何解决

[imchao9]

改为reasoning_effort 提示
[2026-01-22 00:52:33] [76ba02b2] [debug] [conductor.go:2240] Use API key csk-...tcwe for model zai-glm-4.7
[2026-01-22 00:52:33] [--------] [debug] [apply.go:285] thinking: applying config for user-defined model (skip validation) provider=openai model=zai-glm-4.7 mode=level budget=0 level=high
[2026-01-22 00:52:34] [--------] [debug] [openai_compat_executor.go:240] request error, error status: 422, error body: {"message":"Reasoning effort is not supported for this model","type":"invalid_request_error","param":"reasoning_effort","code":"wrong_api_format"}

[luispater]

官方的吗？官方的话，直接用anthropic的协议，走claude-api-key配置项目，然后zai-glm-4.7(high)这样就可以了

[imchao9]

大佬你好，他是 cerebras
https://inference-docs.cerebras.ai/quickstart

他没有支持anthropic 协议

[imchao9]

#1151
我这边改了一下这块，大佬可以看看