Skip to main content
POST /v1/alpha/search is a specialized upstream endpoint, not a uniform provider search API or the Responses web_search tool. This path is restricted to specific channel types; ask your administrator to confirm channel support and configure an available upstream and model. The gateway requires a nonempty model, preserves other raw JSON fields, rewrites model for model mapping, and may apply channel parameter overrides. The source defines no stable query, results, or citation schema. This documentation does not invent one or claim compatibility with an official public OpenAI API. Set up the REST runtime, then obtain a request file matching the deployed upstream version:
The file needs model and the upstream’s search fields. The example rejects stream: true, makes no automatic retries, and checks only JSON object HTTP responses and common error envelopes. Its receipt does not prove semantic search success: validate results, citations, and business error codes against the provider schema. The gateway returns successful upstream bodies and Content-Type without a Chat / Responses wrapper; non-JSON responses are outside this example. This does not promise SSE, uniform usage, or free searches. See the API reference.
阅读中文版