Amux AI

Web Search

Last updated November 22, 2026

Unified web search endpoint routing by model ID and returning structured results with optional full-page content.

Unified synchronous web search endpoint routing by model (currently supporting Jina Search and DeepSeek Search) and normalizing vendor-specific request headers and responses into a consistent JSON format.

POSThttps://gateway.amux.ai/v1/tools/search

Authorization

header
AuthorizationstringRequired

Bearer <your Amux AI key> Amux AI API key created in the console.

Content-TypestringRequiredDefault "application/json"

Must be application/json.

Request

application/json
modelstringRequired

Search model canonical ID or alias (for example jina/jina-search or deepseek/deepseek-search).

querystringRequired

Search query string.

countinteger

Desired number of search results (1–50).

need_contentbooleanDefault true

Whether to include full page content in each result; set false for snippets only.

ori_questionstring

Optional original user question context used for semantic reranking (supported by DeepSeek Search).

localizationstring

Optional client IP string for localized search ranking (supported by DeepSeek Search).

sitestring | array<string>

Optional domain filter (single domain or list of domains).

countrystring

Optional two-letter country code (for example US).

languagestring

Optional language code (for example en).

no_cachebooleanDefault false

Whether to bypass upstream cache for fresh results.

token_budgetinteger

Optional maximum output token budget.

Response

200response

Search succeeded.

400response

Invalid request body or model does not support search.

402response

Insufficient workspace balance.

Protocol overview

  • Unified JSON contract: Switch search providers by changing model without rewriting headers or response parsers.
  • Optional full content: Set need_content: false to return lightweight titles, URLs, and snippets instead of full extracted page bodies.
  • Synchronous, non-streaming: stream: true is rejected with 400; response payloads are capped at 16 MB.

Error handling

Errors follow the standard OpenAI error schema. Missing query, out-of-range count, or unsupported models return 400 invalid_request. See Errors and retries.

cURL
curl https://gateway.amux.ai/v1/tools/search \
  -H "Authorization: Bearer $AMUX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "jina/jina-search",
    "query": "latest AI embedding models",
    "count": 3,
    "need_content": false
  }'
{
  "object": "list",
  "model": "jina/jina-search",
  "results": [
    {
      "title": "Example Result",
      "url": "https://example.com",
      "snippet": "Concise summary of the page.",
      "summary": "<string>",
      "content": "# Example Result",
      "site_name": "<string>",
      "authority": 0,
      "rerank_score": 0,
      "date": "<string>",
      "published_time": "<string>"
    }
  ],
  "usage": {
    "total_tokens": 10000
  }
}