How long does it take?

Usually under a minute for lite and a few minutes for deep, depending on how fast the sources load. Poll the task or use the webhook; there's no long-held HTTP request.

Claude Haiku 5.5 (Anthropic) plans the queries, reviews the gaps and writes the report; OpenAI's Decisions API checks coverage and source relevance. result.llm reports the tokens used. Neither provider trains on API data, and SerpKite never logs the prompts or page text.

Can I see what it searched?

Yes. result.queries lists every Google query, and each source records the query that found it.

How many can run at once?

Up to 20 research and crawl tasks per account can be queued or running at the same time.

POST /v1/research

Deep Research API

Q: What does a report cost?

lite 30, standard 75 and deep 200 credits, flat: $75.00 → $45.00 per 1,000 standard reports, depending on the pack. The price is reserved when the task is queued and charged only when the report is written. A failed task is refunded; canceling mid-run charges 1 credit per search already run.

Hand over a research question. Claude plans Google searches, the pages are read and ranked, gaps get follow-up searches, and you get a Markdown report where every claim cites a numbered source. A flat price per effort level, charged only when the report is written.

Get a free API key Read the reference

75 credits per report at effort=standard (30 lite, 200 deep); charged only when it's written 2,500 free credits, no card Failed calls are free

Overview

What the Deep Research API does

POST /v1/research returns 202 with a task. Claude Haiku 5.5 plans the first round of Google queries; the pages they find are read and ranked against the task; a decision model (OpenAI's Decisions API) checks whether the notes already cover it and drops off-topic sources; Claude reviews what is still missing and plans follow-up rounds. lite runs 3 queries in 1 round, standard 5 queries over 2 rounds, deep 8 queries over 3 rounds.

The result holds the Markdown report with [n] citations, citations, every source with the query that found it, every query it ran and stats. With output_schema it also returns a JSON value in output. Poll GET /v1/research/{id} (free; progress shows the round and step) or take the signed research.completed webhook. Results are kept for 24 hours.

lite costs 30 credits, standard 75 and deep 200, flat. The price is reserved when the task is queued and charged only when the report is written; a task that fails is refunded.

What you get

id string: The task's ID. poll_url is https://api.serpkite.com/v1/research/{id}.
status string: queued, running, completed, failed or canceled.
credits_reserved / credits_used number: The flat price reserved at enqueue, and what was charged when the task ended.
progress object: While running: round, rounds, queries, sources, step (searching, reviewing, writing).
result object: When it ends: input, effort, report (Markdown with [n] markers), citations[], sources[] (index, title, link, domain, published_at, query), queries[], stats (searches, sources, rounds), llm, and output / output_error with output_schema.
error object | null: Set when the task failed (refunded) or was canceled.

Request

POST /v1/research

cURL

Python

TypeScript

fetch

PHP

curl https://api.serpkite.com/v1/research \
  -H "Authorization: Bearer $SERPKITE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"input":"compare espresso machines under $500 for a small office","effort":"standard"}'
# → 202 {"id": "…", "status": "queued", …}; poll until status is completed, failed or canceled:
curl https://api.serpkite.com/v1/research/$TASK_ID \
  -H "Authorization: Bearer $SERPKITE_API_KEY"

Response (illustrative, placeholder domains)

GET /v1/research/{id} · 200 OK (abridged)

{
  "id": "01926a3e-8c4d-7e1f-a2b3-c4d5e6f7a8b9",
  "kind": "research",
  "status": "completed",
  "created_at": "2026-10-24T12:00:00Z",
  "started_at": "2026-10-24T12:00:01Z",
  "completed_at": "2026-10-24T12:01:38Z",
  "credits_reserved": 75,
  "credits_used": 75,
  "webhook_status": null,
  "progress": {
    "round": 2,
    "rounds": 2,
    "queries": 8,
    "sources": 14,
    "step": "writing"
  },
  "result": {
    "input": "compare espresso machines under $500 for a small office",
    "effort": "standard",
    "report": "## Summary\n\nFor a small office, a heat-exchanger machine with a 2-litre tank is the safest pick under $500 [1][3]…",
    "citations": [
      {
        "index": 1,
        "title": "The best espresso machines under $500",
        "link": "https://www.example.com/best-espresso-under-500",
        "domain": "example.com"
      }
    ],
    "sources": [
      {
        "index": 1,
        "title": "The best espresso machines under $500",
        "link": "https://www.example.com/best-espresso-under-500",
        "domain": "example.com",
        "query": "best espresso machine under $500 office"
      }
    ],
    "queries": [
      "best espresso machine under $500 office",
      "office espresso machine tank size"
    ],
    "stats": {
      "searches": 8,
      "sources": 14,
      "rounds": 2
    },
    "llm": {
      "provider": "anthropic",
      "model": "claude-haiku-5-5",
      "input_tokens": 41200,
      "output_tokens": 1650
    }
  },
  "error": null
}

Every response also carries X-Credits-Used, X-Credits-Remaining, X-Cache and X-Tokens-Estimate headers. Full endpoint reference

Reference

Parameters

The fields /v1/research accepts in its JSON body.

Deep Research API parameters
Name	Type	Default	Description
`input` required	string	–	The research task in plain words, up to 8,000 characters (q is accepted as an alias).
`effort`	string	standard	How much to search and read: lite 30, standard 75, deep 200 credits. Values: lite, standard, deep.
`output_schema`	object	–	A JSON Schema (at most 16 KB); the report is also returned as a JSON value in result.output.
`country`	string	–	Search from this country (two-letter code).
`language`	string	–	Search in this language.
`include_domains`	string[]	–	Search only these domains (up to 20).
`exclude_domains`	string[]	–	Never use these domains (up to 20).
`webhook_url`	string	–	Receives the signed research.completed event. Defaults to the account webhook URL.

Full reference, error codes and headers are in the /v1/research reference.

Use cases

What people build with the Deep Research API

Market and vendor research

Comparisons with tables and citations your team can check.

Background agent step

Start a task, keep working, and pick up the report from the webhook.

Enrichment to JSON

output_schema turns a research question into typed fields for your database.

Briefs from trusted sources

include_domains keeps it to filings, official sites or the press you trust.

Pricing