Overview
Endpoint:POST https://api.exa.ai/search. News search is integrated into the main search endpoint. No category parameter needed.
What it searches: Real-time index of web news sources including major publications, trade press, and niche outlets. Semantic search returns results ranked by topical relevance.
Minimal Working Example
Supported Parameters
News search is integrated into the main search endpoint. All standard search parameters are supported:| Parameter | Type | Notes |
|---|---|---|
query | string | Natural language. Topic, company, person, or event. |
type | string | "auto" recommended. All search types supported. |
numResults | integer | 1–100. Default 10. |
includeDomains | string[] | Restrict to specific publications (e.g. ["reuters.com", "techcrunch.com"]). |
excludeDomains | string[] | Exclude specific sources. |
startPublishedDate | string | ISO 8601. Limits to recent articles. |
endPublishedDate | string | ISO 8601. Upper bound on publication date. |
contents | object | text, highlights, summary, all nested under contents. |
Query Patterns
Industry news:Common Mistakes
| Wrong | Correct |
|---|---|
Vague queries like "news" | Be specific: "AI regulation updates in the European Union". The more specific the query, the better the news results. |
Patterns and Gotchas
- News search is integrated into the main search endpoint. No category parameter needed. Just use descriptive news-related queries with
type: "auto". includeDomainscontrols source quality. For trusted sources, restrict to["reuters.com", "bbc.com", "nytimes.com"]. For trade press, use["techcrunch.com", "theverge.com"].- Use
highlightsfor agent workflows. News articles are verbose. Highlights extract the key facts and quotes. - Python SDK uses snake_case.
numResults→num_results,maxCharacters→max_characters. - News works well with deep search. Use
type: "deep"withoutputSchemato extract structured event summaries, sentiment, or entity mentions from news results.

