Skip to main content
Valyu gives one search interface over state-of-the-art web search plus specialised, proprietary, and open data sources spanning research, finance, life sciences, genomics, law, and more. This page is a tool manifest: a single place for an AI agent (or its developer) to learn what exists and exactly how to target it. You have three ways to control which sources answer a query:

Let us route

Call /search with no source filter and Valyu selects the most relevant sources automatically.

Target by id

Pass dataset ids (e.g. valyu/valyu-sec-filings) to included_sources to query exactly those datasets.

Target by preset

Pass a preset name (e.g. finance, genomics) to included_sources to query a whole family at once.
Integrating raw Search? If you are building knowledge work on top of these sources, consider DeepResearch instead - a cost-effective autonomous agent built on top of the same search engine that plans, runs, verifies, and writes a cited report for you. It uses every source below (including DeepResearch-only signals) without you orchestrating the calls.

How to target sources

The included_sources array on /search (and the underlying tooling in Answer and DeepResearch) accepts any mix of:
dataset id
string
A specific dataset, e.g. valyu/valyu-patents. Combine multiple ids to query several datasets together.
preset name
string
A family shortcut: academic, finance, patent, transportation, politics, legal, health, genomics, chemistry, physics, pulse, cybersecurity, environment, automotive, compliance, medical. Expands to every dataset in that family.
collection:NAME
string
An org-scoped saved source set. Create reusable bundles of sources in the platform, then target them by name, e.g. collection:diligence-stack.
web
string
The live web. Pass web to force web search alongside (or instead of) proprietary datasets.
url / domain
string
A specific URL or domain to bias web results toward, e.g. arxiv.org.
Use excluded_sources to remove a source, and source_biases (an object of source -> -5..+5) to soft-rank without hard filtering.

How access works

All plans include web search and open academic sources (arXiv, PubMed). Specialised and proprietary sources (SEC, patents, drug discovery, genomics, and the rest) unlock as you move up plans - and a plan also lowers your cost per credit. During the free trial every source is available so you can test before you commit. See pricing for the full source-to-plan breakdown, or request access to a specific dataset above your plan.
Two availability concepts to know:
  • Plan-gated - a source unlocks at a given plan (or with free trial credits). Marked below as the plan that unlocks it.
  • Access mode - whether a source answers raw /search (and Answer) calls, marked Search + Answer + DeepResearch, or is served only inside DeepResearch, marked DeepResearch only.
DeepResearch-only sources can’t be called via /v1/search. Passing their ids to included_sources on /search returns no results. To use them, run a DeepResearch task (POST /v1/deepresearch/tasks) with the topic in the query; the agent pulls these signals automatically, and you can name the source to nudge it (e.g. “…using OFAC SDN and UK HMT sanctions lists”). Every other source below answers raw Search, Answer, and DeepResearch.
Whole families (every source) - not callable via /v1/search:
  • Transportation - UK National Rail, global ship tracking (AIS)
  • Environment & geo - NWS Weather, OpenAQ, FEMA Disasters, NASA EONET, Open-Meteo (+ air, marine), OSM Nominatim
  • Automotive - NHTSA vPIC, Recalls, Complaints, Investigations
  • Compliance & sanctions - OFAC SDN / non-SDN, UN, UK HMT, Swiss SECO, Australia DFAT, INTERPOL Red, US CSL
  • Pulse signals - GDELT, Wikipedia Pageviews, Crossref Velocity, PyPI / npm / crates.io downloads
Specific datasets inside otherwise-searchable families:
  • Finance - short-seller reports, buyside letters & pitches, semiconductor-industry news, congressional research (CRS), legal & regulatory news, FDIC BankFind
  • Health - NPI registry, WHO ICD, openFDA device events, openFDA food recalls, CDC wastewater (NWSS)
Everything else - web, academic, SEC, patents, chemistry, genomics, economics, prediction markets, UK case law / legislation / Parliament, physics, cybersecurity, and the searchable health/finance datasets - answers raw Search and Answer too.

Catalog

Sizes and identifiers are listed in full on Data Coverage. Below is the agent-targeting view: ids, the preset that selects each family, and the plan that unlocks it.

Web Search + Answer + DeepResearch

Live web search across news, docs, and the open internet. Combine web with any proprietary ids to blend fresh web context into a domain search.

Academic Search + Answer + DeepResearch

Preset: academic • See Academic use cases

Finance & markets Search + Answer + DeepResearch

Preset: finance • See Finance use cases Investment-research signals available inside DeepResearch only (not callable via raw /search): valyu/valyu-short-seller-reports, valyu/valyu-buyside-letters-pitches, valyu/valyu-semiconductor-industry-news, valyu/valyu-congressional-research, valyu/valyu-legal-regulatory-news, valyu/valyu-fdic-bankfind DeepResearch only

SEC filings Search + Answer + DeepResearch

See Finance use cases

Patents Search + Answer + DeepResearch

Preset: patent • See Finance use cases

Chemistry & drug discovery Search + Answer + DeepResearch

Preset: chemistry • See Healthcare use cases

Genomics Search + Answer + DeepResearch

Preset: genomics • See Healthcare use cases 19 NCBI databases. Use the genomics preset to query all at once, or target individual ids: valyu/valyu-ncbi-gene, valyu/valyu-ncbi-nucleotide, valyu/valyu-ncbi-protein, valyu/valyu-ncbi-snp, valyu/valyu-ncbi-clinvar, valyu/valyu-ncbi-geo, valyu/valyu-ncbi-taxonomy, valyu/valyu-ncbi-structure, valyu/valyu-ncbi-sra, valyu/valyu-ncbi-biosample, valyu/valyu-ncbi-genome, valyu/valyu-ncbi-assembly, valyu/valyu-ncbi-bioproject, valyu/valyu-ncbi-cdd, valyu/valyu-ncbi-dbgap, valyu/valyu-ncbi-dbvar, valyu/valyu-ncbi-medgen, valyu/valyu-ncbi-gtr, valyu/valyu-ncbi-protclust. Unlocks on Serious Business.

Health Mixed access

Preset: health • See Healthcare use cases Most health sources answer raw Search; a few registries and recall feeds are DeepResearch-only.

Economic data Search + Answer + DeepResearch

See Finance use cases Preset: legal Legal & regulatory news (valyu/valyu-legal-regulatory-news) is DeepResearch only and listed under Finance above.

Politics Search + Answer + DeepResearch

Preset: politics Congressional Research Service / CRS (valyu/valyu-congressional-research) is DeepResearch only and listed under Finance above.

Transportation DeepResearch only

Preset: transportation • See Data Coverage
These sources are available through DeepResearch, not the Search/Answer API. Run a DeepResearch task with your transportation question in the query.

Physics Search + Answer + DeepResearch

Preset: physics

Cybersecurity Search + Answer + DeepResearch

See Data Coverage

Compliance & sanctions DeepResearch only

See Data Coverage
These sources are available through DeepResearch, not the Search/Answer API. Run a DeepResearch task and name the lists you need in the query (e.g. “screen Acme Corp against OFAC SDN, UN, and UK HMT sanctions”).

Environment & geo DeepResearch only

See Data Coverage
These sources are available through DeepResearch, not the Search/Answer API. Run a DeepResearch task with your weather, air-quality, disaster, or geocoding question in the query.

Automotive DeepResearch only

See Data Coverage
These sources are available through DeepResearch, not the Search/Answer API. Run a DeepResearch task with your vehicle, recall, or complaint question in the query.

Pulse signals DeepResearch only

See Data Coverage
These sources are available through DeepResearch, not the Search/Answer API. Run a DeepResearch task and name the signal in the query (e.g. “PyPI and npm download trends for vector-database clients over the last 12 months”).

Discover sources at runtime

Agents should not hardcode this list. Query the Datasources API to discover what is available, with example_queries (optimised for few-shot prompting), response_schema, and pricing.cpm per source:
Filter to a category with ?category=markets, or list categories with counts at /v1/datasources/categories. This turns Valyu from “an API you need to know” into “an API that teaches itself to your agent.”

Targeting DeepResearch-only sources

Sources badged DeepResearch only are not callable from /v1/search - passing their ids to included_sources returns no results. Put the topic in a DeepResearch task instead and, optionally, name the source to steer it. The access mode lives in each dataset’s access_modes field from /v1/datasources, so agents can check it at runtime.

Teach your agent to discover and target Valyu data sources correctly.

Open in Cursor

Data Coverage

Full registry with sizes and update frequencies

Search API

Use these sources with the Search endpoint

DeepResearch

Autonomous agent over every source

Pricing

Which plan unlocks which sources