Skip to main content

Documentation Index

Fetch the complete documentation index at: https://lava.so/docs/llms.txt

Use this file to discover all available pages before exploring further.

Lava’s first-party fan-out search aggregator: one HTTP call across 9 brokered providers (Tavily, Exa, Jina, Parallel, Perplexity Sonar, Firecrawl, Lava News, Serper, Brave) returning a unified normalized result set in /sources, or an LLM-synthesized answer with citations in /answer (sources are not echoed back — call /sources separately if you need them). Best for agents that need broad cross-source coverage in a single round-trip with explicit partial-failure surfacing. Charges upstream costs plus a 0.005markupon/sources,ora0.005 markup on `/sources`, or a 0.03 markup plus LLM token cost on /answer. 2 endpoints available through Lava’s AI Gateway. See the Lava Search API docs for full documentation.
Supports both managed (Lava’s API keys) and unmanaged (bring your own credentials) mode.

Endpoints

Fan out a query across the 9 default providers (or a caller-specified subset) and return aggregated normalized results plus per-provider failure surfacing. Charges upstream provider costs plus a $0.005 markup.

POST https://api.lava.so/v1/search/sources — $0.005 / request
const data = await lava.gateway('https://api.lava.so/v1/search/sources', { body: {"q":"GLP-1 cardiovascular outcomes 2026 meta-analyses"} });

Same fan-out as /sources plus an LLM synthesis pass — returns { answer, citations, cost } (sources are not echoed back; call /sources separately if you need them). Charges $0.03 markup plus LLM token cost plus upstream fanout costs.

POST https://api.lava.so/v1/search/answer — $0.03 / request
const data = await lava.gateway('https://api.lava.so/v1/search/answer', {
  body: {
"q": "Halo2 vs PLONK vs Spartan tradeoffs for zk-SNARKs without trusted setup"
},
});

Next Steps

All Providers

Browse all supported AI providers

Forward Proxy

Learn how to construct proxy URLs and authenticate requests