Show which model/version was used for each response in Agent mode (especially in Auto mode)

Feature Request: Display the exact model and version used per response turn in Agent conversations


Problem

When using Cursor’s Agent in Auto mode, the model selection happens automatically behind the scenes. As a user, I have no visibility into which model actually generated a given response. This creates a few pain points:

  • Inconsistent confidence: A response from Claude Opus 4 carries different weight than one from a smaller/faster model. Without knowing which was used, it’s hard to calibrate how much to trust the output.

  • Debugging difficulty: When an agent response is unexpectedly poor or surprisingly good, there’s no way to correlate it back to the model that produced it.

  • Reproducibility: When sharing agent outputs with teammates, there’s no record of the model context.

Proposed Solution

Display the model name and version used for each individual response turn directly in the Agent chat UI. For example:

✦ Agent claude-opus-4.5 · just now

This could be shown as:

  • A small label/badge next to the response timestamp

  • A tooltip on hover over the agent avatar/icon

  • Expandable metadata below each message

Why This Matters

This is especially important in Auto mode where Cursor dynamically picks the model. Users deserve to know:

  1. Which model answered their question

  2. Whether a degraded response was due to a cheaper/faster fallback model

  3. How to manually override if needed (informed decision)

Request

Please add per-turn model identification in the Agent conversation view. Even a simple text label would be a significant UX improvement for power users who care about output quality and traceability.


Category suggestion: Feature Requests