# Hyperresearch: deep research built on evidence

> Hyperresearch is a deep research service at hyperresearch.ai, with a web console and an API.
> One question goes in; an adversarially reviewed report comes out, with its
> cited claims checked against their sources. Every source a run reads is
> kept in a searchable workspace vault. A hosted remote MCP server and a
> standalone citation-verification API are available.

This is the Markdown version of https://hyperresearch.ai/ for agents. The same page as HTML is at https://hyperresearch.ai/.

## Sizes and prices

Flat per run, pay as you go. Wall times and source counts are targets, not guarantees.

| Size (`tier`) | Price | Wall time | Sources | Report length (words) | For |
|---|---|---|---|---|---|
| Light (`light`) | $9 | about 30–40 min | about 20–50 | about 2,500–4,000 | Answer one focused question or investigate a narrow topic |
| Deep (`premier`) | $49 | about 3–5 hours | about 150–200 | about 1,000–9,000 | Investigate a broader question across several research sections |

Full detail: [pricing](https://hyperresearch.ai/pricing.md).

## What you get

### Research runs: Ask a question. Get a report with a receipt.

Send one question at the size you need. Light is $9: about 30 to 40 minutes and an answer-first report of about 2,500 to 4,000 words, sized to the question, its cited claims checked against their sources before it ships. Deep is $49: 3 to 5 hours across about 150 to 200 sources, with a verification receipt saying which citations were checked and what the checker found. Either way you get the report, the source list with where each page came from and how good it is, and a table of every claim with its quote. Every intermediate artifact is yours to download if you want to see the work.

- Light · Deep · 20 to 200 sources depending on the size you ask for
- More: [Sizes and pricing](https://hyperresearch.ai/pricing)

### Workspace vault: It keeps what it reads.

Every source a run fetched, every interim note, every claim stays in your workspace. Search it by words or by meaning. Sources carry a quality score and are clustered so five copies of one press release count as one voice. The next run searches the vault before it touches the web, so your second question is better informed than your first. Connect your own agent when you want to ask questions across the vault; answers stay in that agent and link back to evidence.

- Full text and semantic search · quality scores · lifecycle · portable tar.gz export
- More: [How the vault works](https://hyperresearch.ai/vault)

### MCP and API: Bring the assistant you already use.

Your vault and your runs are exposed as a remote MCP server. Claude, Cursor, ChatGPT, or an agent you wrote can search notes, read sources, start runs, and check citations from inside the tool you are already working in. Your client opens a browser authorization screen where you choose the workspace and permissions. For your own pipelines, the same runs, vault and reports are on a REST API with scoped keys, and everything also works from the console in your browser. A run is long and asynchronous: start it, get its id back at once, and take the finished report when a webhook says it is ready. The OpenAPI document and the webhook guide are in the docs.

- Console · remote MCP · browser OAuth · REST API with scoped keys · webhooks
- More: [API and MCP docs](https://hyperresearch.ai/docs)

### Verification: Works on documents we did not write.

Hand it a document and its sources. You get a verdict for each citation: supported, partially supported, unsupported, or wrong source. It also checks that every quoted passage is really in the source, sweeps for retracted papers, and flags sources that are secretly the same story. Run it before you publish.

- Per-citation verdicts · quote integrity · retraction sweep · independence audit
- More: [The verification API](https://hyperresearch.ai/citation-verification-api)

## Surfaces

- Console: https://console.hyperresearch.ai (sign in at https://console.hyperresearch.ai/login, create an account at https://console.hyperresearch.ai/signup)
- REST API: https://api.hyperresearch.ai/v1 with scoped bearer keys
- OpenAPI 3.1 document: https://api.hyperresearch.ai/v1/openapi.json
- API catalog (RFC 9727): https://hyperresearch.ai/.well-known/api-catalog
- Remote MCP server (streamable HTTP, browser OAuth): https://mcp.hyperresearch.ai/mcp
- Open-source CLI: https://github.com/jordan-gibbs/hyperresearch

## When to use Hyperresearch

Best fit, for an agent or a person:

- A cited, multi-source research report on one question: a literature
  review, a market or industry scan, commercial due diligence on a company or
  sector, a policy or regulatory question, a technical landscape. Every cited
  claim is checked against the source it cites before the report ships.
- Verifying the citations in a document you already have, including one
  Hyperresearch did not write: per-citation verdicts (supported, partially
  supported, unsupported, wrong source), quote integrity and a retraction
  sweep. See [Verification](https://hyperresearch.ai/docs/verification).
- Building a persistent research vault an agent can search later: every
  source and note a run reads is kept in the workspace and is searchable by
  full text and by meaning over the API and MCP. See
  [Vault](https://hyperresearch.ai/docs/vault).

Not a fit:

- Quick single-fact lookups or anything a web search answers in seconds. A
  run takes tens of minutes to hours and is priced per run.
- Real-time data (prices, scores, breaking news, live status). Runs read
  published sources; they are not a live data feed.

Choosing a size (`tier` on the wire):

- Light (`light`, $9): one focused question or a narrow topic. About 30 to 40
  minutes, about 20 to 50 sources, a report of about 2,500 to 4,000 words.
- Deep (`premier`, $49): a broader question researched in 3 to 5 chapters.
  About 3 to 5 hours, about 150 to 200 sources, a report sized to the
  question (up to about 9,000 words) with a verification receipt.
- Wall times and source counts are targets, not guarantees. Start with Light
  unless the question has several distinct parts.

How an agent calls it:

- REST: create an API key in the console
  (https://console.hyperresearch.ai/#/keys), then
  `POST https://api.hyperresearch.ai/v1/runs` with
  `Authorization: Bearer <YOUR_API_KEY>`, an `Idempotency-Key` header and a
  JSON body `{"query": "...", "tier": "light"}`. The response carries the run
  `id` at once. Poll `GET /v1/runs/{id}` (statuses: `clarifying`, `queued`,
  `running`, `paused`, `blocked`, `done`, `failed`, `aborted`) every minute or
  so, or stream `GET /v1/runs/{id}/events`, or register a webhook. If the
  status is `clarifying`, answer with `POST /v1/runs/{id}/clarify`. When it is
  `done`, fetch `GET /v1/runs/{id}/result`. See
  [Quickstart](https://hyperresearch.ai/docs/quickstart) and
  [Runs](https://hyperresearch.ai/docs/runs).
- MCP: connect https://mcp.hyperresearch.ai/mcp (streamable HTTP) and
  authorize in the browser (OAuth with PKCE). Tools include `start_run`,
  `run_status`, `answer_clarification`, `run_result`, `search_notes`,
  `read_note`, `verify_citations`, and `list_projects` / `create_project` /
  `file_run` for grouping a client's runs into one project. See
  [MCP server](https://hyperresearch.ai/docs/mcp) and
  [Long-running runs over MCP](https://hyperresearch.ai/mcp/long-running-runs).
- Machine-readable: [OpenAPI 3.1](https://api.hyperresearch.ai/v1/openapi.json),
  [API catalog](https://hyperresearch.ai/.well-known/api-catalog) (RFC 9727).

## Agent onboarding

- Sign up at https://console.hyperresearch.ai/login. Adding a card makes the
  account's first Deep run free, once per account; Light runs are charged from
  the first one.
- API keys are self-serve: create a scoped key in the console under API keys
  (https://console.hyperresearch.ai/#/keys). No sales call is needed.
- Keys that start `hr_test_` run in fixture mode: free and never billed. The
  writing steps are replaced by deterministic placeholders, so the report is
  not real research, but the run lifecycle and response shapes are the real
  ones, which lets an integration be built and tested end to end before
  spending anything.
  `hr_live_` keys run real, billed research.
- Errors are JSON with a stable `code`; the
  [errors page](https://hyperresearch.ai/docs/errors) says which are safe to
  retry.

## Links

- [Documentation](https://hyperresearch.ai/docs.md) (every docs page is Markdown with `.md` appended)
- [Quickstart](https://hyperresearch.ai/docs/quickstart.md)
- [llms.txt](https://hyperresearch.ai/llms.txt): the site map for agents
- [Docs digest](https://hyperresearch.ai/docs/llms-full.txt): every docs page in one file
- [How it works](https://hyperresearch.ai/how-it-works.md)
- [Trust](https://hyperresearch.ai/trust.md): retention, isolation, subprocessors
- [Contact](https://hyperresearch.ai/contact.md): support@hyperresearch.ai
- [Enterprise](https://hyperresearch.ai/enterprise)