List models
GET/v1/models
Live-query every registered provider for its available models and
return a best-effort merged list. A provider that fails discovery, or
takes longer than 5 seconds, is left out of the otherwise successful
200 response. Entries are not deduplicated. On the Enterprise platform,
model-access rules narrow the list to the models the workspace may use.
Streaming, vision, tool-call, structured-output and JSON-mode flags are
the provider's declared capabilities. max_context_tokens and
max_output_tokens are per model, from a model_limits entry or the
model catalogue, where one is known. The public object does not expose
batch or streamed-tool-call support.
When the request carries an anthropic-version header, as Claude Code
sends, the answer is in Anthropic's list shape and holds Claude models
only. A key with scopes needs models:read.
Request
Responses
- 200
- 401
- 403
- 429
Model list, in the OpenAI shape, or in Anthropic's shape when the request carries anthropic-version.
No key was sent (api_key_required), or the key is unknown (invalid_api_key), revoked (api_key_revoked) or expired (api_key_expired).
The key's nonempty scope list does not include models:read (api_key_scope).
The key's rate limit is reached (rate_limit_exceeded, with a Retry-After header), priority admission control throttled the request (priority_throttled), or the provider is rate limiting the call (provider_rate_limited, with Retry-After when the provider gave one).
Response Headers
Seconds to wait before retrying, when known.