Models
Find available models and inspect their metadata. Provider-compatible model lists are grouped with inference models. List provider-compatible inference models. Read, update, or reset project model access: the allowed models, the denied models and the default model. Requires project.settings.manage. Read and update the models an account allows and denies and its default model, which every project of the account inherits. These endpoints require a signed-in account owner or admin.Provider credentials
Configure account-level Anthropic and OpenAI keys, preview routing restrictions, or revoke credentials. Requires a signed-in account owner or admin.Inference
Create Anthropic-compatible messages. Count input tokens without generating content or charging credits when the selected Anthropic deployment supports token counting. Create OpenAI-compatible chat completions. Create OpenAI-compatible responses. Create embeddings. Call Gemini-compatible generateContent methods. Generate images from a prompt. Manage generated images through project upload endpoints. Use billing finalization when your integration manages gateway billing groups. Inspect spending limits through budgets. Read routing or access metadata for proxy requests. These endpoints have separate freshness and authorization requirements.Model usage
Read recorded model calls and token consumption by account or project, optionally for one member. Both reports return model totals and daily model breakdowns; the account report also breaks usage down by project.
Back to REST API reference