Documentation
Vaaya API reference
One vaaya_sk_ key unlocks the whole catalog — data, LLMs, media, search, scraping, compute, and storage — over plain HTTP, the OpenAI-compatible LLM router, and MCP. No per-vendor keys, no subscriptions. Pay per call; failed calls are never charged.
One call shape
Every catalog action is the same endpoint: POST /api/run/{service}/{action} with the action’s params as the JSON body and your key as a bearer token. Every response uses one envelope, and billing happens only when the upstream call succeeds.
curl -X POST https://vaaya.ai/api/run/exa/search \
-H "Authorization: Bearer $VAAYA_API_KEY" \
-H "Content-Type: application/json" \
-d '{"query": "latest developments in WebTransport", "numResults": 10}'Don’t know which service you need? POST /api/run/vaaya/consult (1¢) turns a plain-English intent into the exact calls to make, and POST /api/run/vaaya/discover (free) keyword-searches the long tail of thousands of endpoints.
The services
Each service page in the left nav explains what the service is good at, what each action costs, and the nuances that matter in practice — which vendor to default to, when to escalate to a pricier tier, and the gotchas that waste money or return stale data.
| Category | What it covers |
|---|---|
| Web search | Five search vendors — semantic, independent-index, RAG-tuned, cited answers — from 1¢ per query. |
| SuperSearch | One flat-price call that routes, races, corroborates, and extracts across the search stack for you. |
| Web scraping | Content extraction, JS rendering, crawls, structured extraction, and anti-bot escalation. |
| Browser | Real cloud browser sessions for clicking, typing, and form-filling — acting, not reading. |
| LLMs | 300+ models behind one OpenAI-compatible endpoint, metered per token. |
| Media | Image, video, audio, and speech generation — the model is a parameter. |
| Compute | Ephemeral sandboxes and GPU boxes for running code in isolation. |
| Files & storage | Durable, billed-once file storage your agent can write and share. |
| Memory | Long-term agent memory across sessions — three providers, one interface. |
| Inboxes your agent owns: send, receive, and act on replies. | |
| GTM & leads | Find, enrich, and verify prospects; signals and outreach plumbing. |
| Research | Managed multi-step research runs that produce cited reports. |
| Public records | SEC filings, court dockets, nonprofit 990s, H-1B salaries — receipts included. |
| Real estate | US property records, valuations, listings, and owner data. |
| Social data | Profiles, posts, and comments across social platforms, including CN platforms. |
| On-chain & markets | Crypto, wallet forensics, and prediction-market data. |
| Compliance & KYB | Sanctions, PEP, adverse media, and company-registry checks. |
| Open data | The long tail of open catalog endpoints, searchable for free. |
Pricing model
- Every action has a fixed or capped price in cents, shown on its service page and in the machine-readable index.
- You are charged only on success — an upstream 4xx/5xx charges you nothing.
- Pass
max_cost_centsas a guard on any call: it is a ceiling, not the charge. The response reports the actualcharged_cents. - New accounts get a $2 welcome credit line the moment a card is on file — the card settles only what your agents actually use. Every call is itemized on your transactions.