Multi-source briefs with citations, JSON Schema output, and predictable per-task budgets. Ship features powered by real research without burning a model bill.
curl https://api.infrence.ai/v1/research \
-H 'Authorization: Bearer inf_live_…' \
-H 'Content-Type: application/json' \
-d '{
"question": "Compare the top 5 vector databases by hosted pricing",
"mode": "standard"
}'{
"id": "rsh_2P4sQzDk9wM",
"status": "ready",
"credits_charged": 60,
"answer": "Pinecone, Weaviate, Qdrant, Milvus, and Chroma cover most of the hosted-vector market. Pinecone starts at $0/mo (Starter), Weaviate Cloud at $25/mo, Qdrant Cloud at $0.014/h …",
"sources": [
{ "title": "Pinecone Pricing", "domain": "pinecone.io" },
{ "title": "Weaviate Cloud", "domain": "weaviate.io" },
{ "title": "Qdrant Cloud", "domain": "qdrant.tech" }
],
"elapsed_ms": 11842
}$0.01 per credit. No subscription, no card on file. Buy a pack, build with it, top up if you run out.
Heavy usage? Subscribe and pay less per credit. Cancel any time.
Pick the mode that fits the job. Lower modes are cheaper and faster.
| Mode | Credits | List price (USD) | Includes |
|---|---|---|---|
| Lite | 15 | $0.15 | 1 wave, 5 sources, ~60s |
| Standard | 60 | $0.60 | 2 waves, 20 sources, ~3 min |
| Pro | 200 | $2.00 | 3 waves + critic, 60 sources, ~6 min |
| Max | metered | your max_credits | Up to your max_credits, 120 sources, ~10 min |
Every call accepts max_credits. We refuse over-budget work or downgrade — no surprise invoices.
Pass a JSON Schema; get a typed object back. Validation failures refund.
Every brief ships with title, URL, domain, and an optional per-source summary.
Long jobs return a 202 + id. Stream progress via SSE or get a signed POST when done.
Restrict the crawler to vetted domains, or banish the ones you do not trust.
DELETE the job. Reserved credits refund automatically.