How is pricing structured?
You buy a monthly plan with included credits, or top up with a one-off pack at the same per-credit rate. 1 credit = $0.01 list price. /v1/research charges a fixed credit price per mode (Lite 15, Standard 60, Pro 200) or metered for Max.
What does 1 credit equal in dollars?
$0.01 list price. The actual margin depends on the LLM / search vendors we use under the hood, but the developer-facing math stays simple.
What are research modes?
Modes pick the budget envelope — wall-clock cap, max waves, max sources. Lite is ~60s and 5 sources. Standard is the default at ~3 minutes and 20 sources. Pro adds a critic pass. Max is metered — you supply max_credits and we charge actual cost.
Can I get a typed object back instead of prose?
Yes — pass a JSON Schema in `response_schema` and the brief is re-cast into a typed object that conforms to the schema. Strict mode is on by default; failed validation refunds your credits.
How do citations work?
Every brief comes with a `sources` array — title, URL, domain, and (optionally) a per-source summary. We track per-claim provenance internally; ask us if you need it surfaced.
Can I run jobs async or with a webhook?
Yes. Long-running modes default to async — you get a 202 with a job id. Pass `webhook_url` to be notified when the brief is ready, or stream progress via the SSE endpoint. Cancel any time.
Are there rate limits?
Every plan has a concurrent-jobs cap and standard request rate limits. Hard caps are listed on each plan tier on the pricing page.
Do you keep my data?
By default we retain request/response bodies for 30 days for debugging. Enterprise customers can opt into Zero Data Retention — nothing is persisted past the response.
Can I bring my own model API key?
Enterprise only. Reach out and we will route the inference through your account.
Refunds?
Failed jobs are refunded automatically. Unused subscription credits roll over within the same plan; pack credits are valid for 6 months.