Claude Code Fixes Subagent Models and Asks Cut-Off Subagents for the Whole Answer Again
Claude Code 2.1.325 stops subagents with their own model from being moved to the plan-mode model when the setting is opusplan or haiku. It also sizes their file reads and MCP results with their own model. Subagents cut off at the output token limit are now asked for the whole answer again, more briefly.
Key Takeaways
- Subagents keep their own model under
opusplanandhaiku. - File reads and MCP results are sized for the subagent's model.
- Cut-off subagents are asked for the whole answer again, more briefly.
- Users no longer get only the tail of an answer.
- A permission mode no longer reverts after a subagent's plan is approved.
- The fixes help sessions that mix models.
Model handling
Subagents with their own model were moved to the plan-mode model when the model setting was opusplan or haiku. That no longer happens. Subagents on a different model also sized file reads and MCP results with the main session's model, and now use their own.
Cut-off answers
A subagent cut off at the output token limit used to return only the end of its answer. It is now asked for the whole answer again, more briefly.
Permission mode
A session sometimes went back to an earlier permission mode after a subagent's plan was approved. That is fixed.