GET /v1/models/{model_id}/health

Part of the API reference; this page is generated from the OpenAPI document that the gateway builds from its own route tables.

Provenance

• Contract version: 1.10.0
• Contract title: acumen gateway API
• Source: acumen:docs/openapi/acumen-v1.openapi.json

Operations

GET /v1/models/{model_id}/health

Live per-model availability

The live availability of one served model, from the same state the door dispatches with: per distinct seat (provider + vendor model, never the account) — live/cooling, cooling-until, in-flight count, cache warmth, token-cap state and last error code. ?probe=true additionally runs one minimal upstream call per seat (verdicts cached 30s) and adds a probe verdict per seat. A known model whose every seat is cooling or over cap answers 503 with Retry-After (soonest recovery); an unknown name is 404 model_not_found.

• Operation id: getV1ModelsModelIdHealth
• Scope: chat
• Exposure: public
• Mutating: no
• Auth: bearerAuth, apiKeyAuth

Responses

• 200: live availability report
  • application/json: ModelHealth: ModelHealth (object)
• 404: model_not_found — OpenAI error envelope
  • application/json: OpenAIError: OpenAIError (object)
• 503: model_no_eligible_seat — every seat cooling or over cap; Retry-After carries the soonest recovery
  • application/json: OpenAIError: OpenAIError (object)