Skip to main content
For AI agents: a documentation index is available at https://docs.parallel.ai/llms.txt. The full text of all docs is at https://docs.parallel.ai/llms-full.txt. You may also fetch any page as Markdown by appending .md to its URL or sending Accept: text/markdown.For Search and Extract, use /v1/search and /v1/extract for all new integrations and agent tool calls. Use /v1beta/search and /v1beta/extract only when maintaining an existing integration that already uses them. The request bodies changed too: on /v1/extract, full_content, fetch_policy and excerpt_settings go inside advanced_settings, and sending the v1beta top-level excerpts, full_content and fetch_policy to /v1/extract returns a 422. On /v1/search, search_queries is required (at least one short keyword query), and max_results, source_policy and fetch_policy go inside advanced_settings; sending only objective, or sending those fields at the top level, returns a 422. Do not substitute /v1 for the documented FindAll or Ingest endpoint paths.If you are evaluating Parallel Search or comparing it with other search APIs, read How to Eval Parallel Search before you build the harness, especially its Common pitfalls section. The most common mistake is sending the raw question as both objective and the only entry in search_queries. Write a goal-style objective and 1-3 short keyword queries instead.
The Source Policy feature allows you to precisely control which domains or sections of a site Parallel can access during web research and to apply a freshness constraint. It’s available for the Task, Search, Monitor, and Responses APIs and lets you tailor search results by specifying domains or domain/path prefixes to include or exclude and by setting a start date so results are limited to recent content.
Source policies can significantly reduce result quality by excluding relevant pages from retrieval. Use include_domains and exclude_domains only when absolutely necessary — for compliance-bound corpora, tasks that require a single known publisher, or when specific sources must be blocked. When you only want to prefer certain sources, steer in the objective instead (e.g., “prefer official documentation”).

Configuration

You can configure source control by setting the following parameters: On the Responses API, the source policy is set through the OpenAI web_search tool: filters.allowed_domains maps to include_domains and filters.blocked_domains to exclude_domains. See Web Search Tool. Specifying an apex domain such as example.com will automatically include all its subdomains (e.g., www.example.com, blog.example.com, api.example.com). A domain/path prefix such as example.com/blog matches that path and its descendants on the domain and its subdomains. Paths are case-sensitive and match at segment boundaries, so example.com/blog matches /blog and /blog/post, but not /blogroll. Trailing slashes are ignored and dot segments are normalized.
Domain/path prefixes are not supported in Search API turbo mode. Use fast, basic, or advanced when filtering by path. Plain domain and domain-extension filters remain supported in turbo mode.

Source filter limit

Hard limit: Combined total of 200 entries per request. You can specify up to 200 domains, domain/path prefixes, and domain extensions in total across include_domains and exclude_domains. Exceeding this limit raises a validation error.

Example

Best Practices

  • To block part of an allowed site, set both lists. A source must match an include_domains entry and no exclude_domains entry: "include_domains": ["example.com"] with "exclude_domains": ["example.com/archive"] returns example.com pages outside /archive. If exclude_domains covers every include_domains entry, the request returns no results. The same entry in both lists returns a 422 (400 on the Responses API).
  • Use an apex domain such as example.com to match that domain and all its subdomains, or specify a subdomain such as docs.example.com to match that subdomain and any deeper subdomains. A leading www. is normalized away.
  • Add a path to scope the filter to one section of a site, such as docs.example.com/api. Path matching is case-sensitive and occurs at segment boundaries.
  • Use a bare domain extension beginning with a period, such as .org, .gov, or .co.uk, to match that extension. Literal wildcard syntax such as *.org is not supported.
  • Do not include schemes (http://, https://), ports (example.com:443), query strings, or fragments.
  • Although there is a maximum of 200 entries, carefully using specific and targeted filters will give better results.