During Droid execution, show a real-time output token counter next to the streaming status, similar to:
streaming ↓ 153
The number should increase live as the model generates tokens.
This is different from showing cumulative token usage after/between model calls. The key requirement is to provide immediate visual feedback while the model is actively streaming.
✶ Detecting… (5s · streaming · ↓ 153 · esc to interrupt)
During Droid execution, show a real-time output token counter next to the streaming status, similar to:
streaming ↓ 153
The number should increase live as the model generates tokens.
This is different from showing cumulative token usage after/between model calls. The key requirement is to provide immediate visual feedback while the model is actively streaming.
✶ Detecting… (5s · streaming · ↓ 153 · esc to interrupt)