pi-mono/scripts/issues/categories/models.md
Armin Ronacher 2e0f175bd5
feat(ai): refactor provider definitions (#11368)
This builds the model catalog from per-provider definitions and
models.dev v2 data.

It eplacer `generate-models.ts` with one `defineProvider()` file per
built-in provider, shared model-family rules (Claude, GPT, DeepSeek,
Kimi, Google), and source adapters. models.dev is read through its v2
API. Outputs and CLI flags are unchanged.
2026-10-28 11:06:10 +01:00

740 B

owner
davidbrai

models

Which models exist and which one is used.

In:

  • Built-in model catalog (packages/ai/catalog, models.dev, pi.dev registry).
  • Model metadata: context window, max tokens, thinking level map, pricing.
  • models.json custom providers and models.
  • Model resolution: --model, --provider, defaultModel, startup fallback, scoped models, virtual models.
  • Contents and ranking of /model, thinking level selection.

Not here:

  • How the model selector is drawn: tui.
  • Wire format and request contents: providers.

Examples: #9566, #9884, #10160, #9099, #10236, #10552, #8810, #10507.

Usually not worth opening: "add model X" when the data comes from models.dev (fix upstream), stale catalog fixed by updating pi.