Skip to main content
Request a model by name with the model field. Each plan includes a set of models and higher plans add more (Plans & limits); in PAYG mode every model is available. Claude Opus, Claude Fable and the older Claude models run in PAYG mode only. Each request takes credits for the model that served it, and the dashboard shows what every request used. Model names link to the full page for that model, with its description, strengths and capabilities. Closed-weight models are called by their name (gpt-6.1-sol). Open-weight models are called @sdn/<name> (@sdn/glm-5.3). The older spellings (sdn-glm-5.3 and the bare glm-5.3) keep working, so configs you already have need no change. In YAML, quote an id that starts with @: model: "@sdn/glm-5.3".

​
Live and coming soon

Every model below is live unless its name is marked Coming soon. A coming-soon model is announced but not served yet: a request for it returns 404 model_unavailable with models you can use instead. Check for the mark before you pin an id in production.
This page is a snapshot. Call GET /v1/models to read the catalog programmatically — each entry is self-describing, down to its capabilities and limits. The endpoint is always authoritative over this page.

​
Text models

Newest models first, then by tier.

​
Embeddings

Turn text into vectors — for semantic search, RAG, clustering, and dedupe.