Deep research

Research is a workflow, not a single call: find sources, corroborate across independent indexes, extract, and cite. This page routes you to the right rung — and documents two primary-source gateways search can’t match: kadec0/fetch (papers, filings, CVEs, government data) and google-trends/fetch (real search-demand curves).

When to use which

Need	Call	Price	Pick when
One-pass “find pages about X”	`exa/search`	1¢	The default — see Web search.
Cited answer in one call	`linkup/search`	1¢	You want an answer with sources, not links.
Route + race + corroborate, one call	`vaaya/supersearch`	5¢	The intent is “answer this well” — see SuperSearch.
Multi-hop agentic retrieval	`linkup/deep-search`	5¢	Only after one-pass search comes back thin.
Full multi-step research run, async	`parallel/task`	10–30¢	Poll `parallel/task-status` free — see Web search.
Extract and persist source pages	scrape + store	from 0.1¢/URL	Web scraping → Files & storage.
Primary-source public data	`kadec0/fetch`	1–11¢	This page — papers, CVE, FDA, SEC EDGAR, congress trades, and more.
Search-demand curves	`google-trends/fetch`	~5¢	This page — quantitative demand data for market research.

Adjacent depth lives on its own pages: Public records for filings, dockets, and 990s with receipts; GTM & leads for company-level intel and news signals; Social data for platform-native data. For evaluative questions — “best X for Y” — don’t stop at rankings: pull candidates with search, then measure on your own data (a sandbox from Compute is the honest way to do it).

kadec0/fetch — primary-source lookups

Price: 1–11¢, capped per endpoint
Shape: GET-style — the endpoint path is a param
Discovery: vaaya/discover — free

One gateway, dozens of primary-source endpoints. The endpoint path goes in the body as endpoint, alongside that endpoint’s own query params. Find the exact slug, price, and required params with vaaya/discover (free) — never guess.

# 1) find the endpoint + its required params (free)
curl -X POST https://vaaya.ai/api/run/vaaya/discover \
  -H "Authorization: Bearer $VAAYA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"query": "congress trades"}'
# → [{ "service": "kadec0", "action": "fetch",
#      "endpoint": "/v1/congress-trades", "price_cents": 7, ... }]

# 2) call it — the hit's params go alongside "endpoint"
curl -X POST https://vaaya.ai/api/run/kadec0/fetch \
  -H "Authorization: Bearer $VAAYA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"endpoint": "/v1/congress-trades"}'

What’s behind it:

Academic — /v1/papers, /v1/paper-fulltext, /v1/related-papers, /v1/author-search.
Government & regulatory — /v1/sec-edgar (7¢), /v1/congress-trades (7¢), /v1/fda + /v1/recalls, /v1/permits, /v1/usgs-quakes.
Security & infrastructure — /v1/cve, /v1/whois-bulk, /v1/domain-info, /v1/dns-propagation, /v1/website-audit (11¢).
Developer & reference — /v1/hackernews, /v1/stackoverflow, /v1/github, /v1/geocode, /v1/weather (1¢), /v1/nutrition.

Gotchas

Cite the primary source in reports — that’s the whole point of going to the gateway instead of a search snippet.
For richer filings work (full-text SEC search, dockets, 990s), see Public records.

google-trends/fetch — search demand

Price: ~5¢, capped
Shape: GET-style — endpoint + keyword

Real Google search-demand data — the quantitative side of market and demand research. Endpoints: /trend (interest over time), /interest-by-region, /related-queries, /related-topics, each taking a keyword.

curl -X POST https://vaaya.ai/api/run/google-trends/fetch \
  -H "Authorization: Bearer $VAAYA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"endpoint": "/related-queries", "keyword": "vector database"}'

Pair it with exa/search for the qualitative side: Trends says how much people want it; search says what they say about it.