Models and aliases
What you can call, and what it costs.
Models in the app lists every model your account can use, with prices per 1M tokens and your own usage. The list follows what's live upstream, so it changes over time. Use the id exactly as shown, for example claude-sonnet-4-5-20250929.
GET /api/v1/models returns the same list in the format your client expects.
Switching models is just a different model in the request, so most tools let you change it without touching your setup. In Claude Code that's /model <id>, see Claude Code.
Aliases
An alias like simpang/fast points at an ordered list of models. When the first one is rate limited or down before any output is sent, Simpang tries the next. Aliases are listed under Models when the admin has set them up.
Access
You see the models your workspace has, minus any a workspace admin turned off for you. A key can be narrowed further to a few models when you create it. Calling anything else returns 403 model_not_allowed with the list of models you can use.
Cost
Each call is priced from its token counts:
| Part | Price |
|---|---|
| Input (uncached) | input price |
| Cache read | cache read price |
| Cache write | cache write price |
| Output (thinking included) | output price |
Cost per part is tokens × price per 1M ÷ 1,000,000, rounded up to the next billionth of a dollar.