Xiaomi MiMo Token Plan Anthropic-Compatible Endpoint: 400 Error with Thinking and Tool Use Replay

What this error means

MiMo Token Plan Anthropic-compatible endpoint returns 400 when thinking + tool_use history is replayed is a Anthropic API failure pattern reported for developers trying to fix 400 error when using xiaomi mimo token plan anthropic-compatible endpoint with thinking mode and tool_use message replay. Based on the imported evidence, treat this as a tool-specific troubleshooting page rather than a generic API error.

Why this happens

Reproducible on macOS Apple Silicon with Craft Agents 0.9.2. MiMo Token Plan is a paid Xiaomi AI service. The endpoint at token-plan-cn.xiaomimimo.com/anthropic implements a subset of Anthropic Messages API. Thinking + tool_use replay triggers 400.

Common causes

• Developers using Xiaomi MiMo Token Plan’s Anthropic-compatible API endpoint encounter 400 errors when the conversation history contains both thinking blocks and tool_use messages. The endpoint implements a subset of the Anthropic Messages API but fails to handle the thinking + tool_use combination correctly, breaking multi-turn agentic workflows.
• Reproducible on macOS Apple Silicon with Craft Agents 0.9.2. MiMo Token Plan is a paid Xiaomi AI service. The endpoint at token-plan-cn.xiaomimimo.com/anthropic implements a subset of Anthropic Messages API. Thinking + tool_use replay triggers 400.

Quick fixes

1. Confirm the exact error signature matches MiMo Token Plan Anthropic-compatible endpoint returns 400 when thinking + tool_use history is replayed.
2. Check the Anthropic API account, local tool state, and provider configuration involved in the failing workflow.
3. Compare the failing environment with a known working setup, then change one configuration value at a time.

Platform/tool-specific checks

• Verify the command, editor, extension, or API client that produced the error.
• Compare local settings with CI, deployment, or editor-level settings when the error appears in only one environment.
• Avoid deleting credentials, local model data, or project settings until the failing scope is clear.

Step-by-step troubleshooting

1. Capture the exact error message and the command, editor action, or request that triggered it.
2. Check whether the failure is account/auth, quota/rate, model/provider, local runtime, or deployment configuration.
3. Review the source evidence below and compare it with your environment.
4. Apply one change at a time and rerun the smallest failing action.
5. Keep the working fix documented for the team or deployment environment.

How to prevent it

• Keep provider/tool configuration documented.
• Record non-secret diagnostics such as tool version, provider name, model name, and command path.
• Add a lightweight check before CI or production workflows depend on the tool.

Sources checked

• https://github.com/craft-ai-agents/craft-agents-oss/issues/746

Evidence note: Reproducible on macOS Apple Silicon with Craft Agents 0.9.2. MiMo Token Plan is a paid Xiaomi AI service. The endpoint at token-plan-cn.xiaomimimo.com/anthropic implements a subset of Anthropic Messages API. Thinking + tool_use replay triggers 400.

Related errors

• Anthropic API thinking block compatibility errors
• Anthropic Messages API tool_use replay failures
• Anthropic-compatible endpoint 400 errors with agentic workflows

FAQ

What should I check first?

Start with the exact MiMo Token Plan Anthropic-compatible endpoint returns 400 when thinking + tool_use history is replayed text and the smallest action that reproduces it.

Can I ignore this error?

No. Treat it as a failed Anthropic API workflow until the root cause is understood.

Is this guaranteed to have one fix?

No. The imported evidence supports the troubleshooting path above, but tool behavior can vary by account, plan, version, provider, and local configuration.

How do I know the fix worked?

Rerun the same command, editor action, or request. The fix is working when that action completes without MiMo Token Plan Anthropic-compatible endpoint returns 400 when thinking + tool_use history is replayed.