Appearance
Web search
POST /v1/search searches the web through one model, usefully/web-search, served by Exa.
bash
curl https://api.usefully.info/v1/search \
-H "Authorization: Bearer $USEFULLY_API_KEY" -H "Content-Type: application/json" \
-d '{"query": "open-source vector databases", "num_results": 5, "freshness": "month"}'| Field | Notes |
|---|---|
query | Required |
num_results | 1–20 (default 10) |
freshness | day, week, month or year |
include_domains / exclude_domains | Restrict or exclude sites |
include_content | Also return page text (content), up to max_content_chars |
provider | Routing preferences, e.g. {"only": ["exa"]} |
Each result has title, url, snippet, and when available published_at, content and score. Results are normalized, so switching engines doesn't change your code.
Search is billed per query, not per token; see the price in GET /v1/models (pricing.per_thousand_requests).