8000 ADK: Missing Thought Process Output with Claude Sonnet 4 Despite Enabled Parameters · Issue #1489 · google/adk-python · GitHub
[go: up one dir, main page]

Skip to content
ADK: Missing Thought Process Output with Claude Sonnet 4 Despite Enabled Parameters #1489
Open
@rick-bm

Description

@rick-bm

Describe the bug
When using the ADK with Claude Sonnet 4 model, I've configured parameters to enable thought process output, but the API response doesn't include any reasoning steps despite the model's documented chain-of-thought capabilities.

To Reproduce

MODEL_CLAUDE_SONNET = "anthropic/claude-sonnet-4-20250514"

root_agent = Agent(
    model=LiteLlm(model=MODEL_CLAUDE_SONNET),
    name='root_agent',
    instruction="""answer user questions""",
    planner=BuiltInPlanner(
        thinking_config=types.ThinkingConfig(
            include_thoughts=True,
        ),
    ),
)

Expected behavior
Response should contain thought_process or similar field with intermediate reasoning steps as per documentation.

Desktop (please complete the following information):

  • OS: MacOS
  • Python version(python -V): 3.13.3
  • ADK version(pip show google-adk): 1.2.1

Additional context
https://docs.anthropic.com/zh-CN/docs/build-with-claude/extended-thinking

Metadata

Metadata

Assignees

Labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions

    0