--context-window | 200000 | Context window in tokens |
--description | | Short description |
--enabled | true | Model is selectable in the picker |
--endpoint | | OpenAI-compatible endpoint id (omit for Ankra-managed Claude) |
--key | | Catalog key (lowercase slug; the model_mode value) |
--max-output | 8192 | Maximum output tokens |
--model-id | | Provider model id (e.g. claude-opus-5 or gpt-4o) |
--name | | Display name shown in the picker |
-o, --output | | Output format: json or yaml (default: human-readable) |
--sort-order | 0 | Sort order in the picker (lower first) |
--supports-images | false | Model accepts image inputs |
--supports-thinking | false | Model supports a thinking/reasoning channel |
--supports-tools | true | Model supports tool calling |
--tier | | Auto-routing tier: expert, think, quick, or empty |