Current behaviour: The status line shows ctx: 15.8k/200.0k (7.9%). The 200.0k denominator is hard-coded regardless of which model is active. This is wrong for most models.
Desired behaviour: The context window size shown should match the actual limit of the model in use.
Guidance:
The Anthropic API exposes a models endpoint:
GET https://api.anthropic.com/v1/models/{model_id}
This returns per-model metadata including max_input_tokens, max_tokens, display_name, and capability flags. For example, claude-opus-4-6 returns max_input_tokens: 1000000 and display_name: "Claude Opus 4.6". These fields may also be useful beyond just the context window fix.
Current behaviour: The status line shows
ctx: 15.8k/200.0k (7.9%). The200.0kdenominator is hard-coded regardless of which model is active. This is wrong for most models.Desired behaviour: The context window size shown should match the actual limit of the model in use.
Guidance:
The Anthropic API exposes a models endpoint:
This returns per-model metadata including
max_input_tokens,max_tokens,display_name, and capability flags. For example, claude-opus-4-6 returnsmax_input_tokens: 1000000anddisplay_name: "Claude Opus 4.6". These fields may also be useful beyond just the context window fix.