Skip to main content

​
Overview

Seltz Search returns ranked, up-to-date results with their contents from an owned index.

​
Search scopes

Search targets a specific scope. A scope is a curated subset of the owned index. Each scope contains a different type of data. Different scopes return data with different contents and structures.

​
Search tiers

Search runs in one of two tiers, selected with the tier parameter. A tier picks which side of the latency / quality trade-off you want. To change which data to search, use scope. The two are independent: any scope can be requested in either tier. Both tiers search the same data and return the same shape of result: only ordering quality and response time differ. Omitting tier gives you pro, so base is opt-out rather than opt-in: send it explicitly when latency matters more than ordering. Every scope accepts either tier.
Tier names are matched case-insensitively and surrounding whitespace is ignored, so pro, PRO, and " Pro " are one tier. An unrecognized name is rejected rather than quietly falling back to the default, so the tier you are served is always the tier you named.

​
Structured for LLM context

Each result is returned as a url and content, with a published_date on time-stamped scopes such as news, rather than raw HTML, so you can pass results directly into a prompt or pipeline without additional cleanup. content is capped at 20,000 characters by default. Use fields.content to raise or lower that ceiling, or turn content off entirely. You can also request snippets for your results. Snippets are short, relevant excerpts from each result. Snippets are useful when you want the most relevant excerpts without passing an entire document into your prompt or pipeline. You can request results that only include snippets or full-page content and snippets, and tune how many snippets you get and how large each one is. Snippets are available on the web, news and wikipedia scopes; companies and people don’t produce them.

​
Search vs. Answer

Search retrieves the citations that /answer uses internally to ground its generated response. Use Search directly when you want the raw sources yourself. For example, use Search to build a custom RAG pipeline or display source documents in your own UI.

​
Next steps

To learn more about Seltz search, check out the following topics:
  • Basic search — Making your first search request
  • Choose a search tier — Trading result quality for lower latency
  • Get snippets — Retrieve a document’s highest-scoring passages
  • Control content length — Raise or lower how many characters of content you get back
  • Scope — Learn about the different scopes available for Search
  • Answer — Learn about Answer and how it uses Search
  • Search API Reference — Full request and response specification
  • Errors — Status codes and error handling for Search and Answer