Expected Behavior

The deepseek-v4-pro model natively supports a 1M token context window. When configuring and using this model within the Command Code agent, it should be able to utilize the full 1M context limit to handle large-scale codebase analysis and reasoning tasks.

Actual Behavior

The context limit for the deepseek-v4-pro model in the agent is currently restricted to 200k tokens. When providing context that exceeds this 200k limit, the agent cannot process it fully, which underutilizes the model's actual capability.

Command Code Version

0.25.1

Operating System

macOS

Terminal/IDE

No response

Shell

No response

Additional context

No response