Agent runs — /v1/agent-runs and everything under it, plus /v1/states and /v1/agent-states | 410 | Run your agent loop in your own client against Chat Completions, Responses or Messages. For long work that should survive a dropped connection, use Responses with background: true. |
Artifacts — /v1/artifacts and everything under it | 410 | Keep large results in your own storage and send the slice you need in the prompt. |
Batches — /v1/messages/batches and /v1/batches | 410 | Send the requests individually (with an Idempotency-Key so retries are safe), or use Responses background: true for long-running items. |
| The Responses WebSocket transport | 410 | POST /v1/responses over HTTPS, with stream: true for Server-Sent Events. |
Legacy text completions — /v1/completions | 410 | Chat Completions: put your prompt in a single user message. |
Tool connectors and catalog — /v1/tool-connectors, /v1/tool-catalog | 410 | Declare only the tools the model itself executes. See Tool calling. |