# Hyperresearch > Hyperresearch is a deep research API and console at hyperresearch.ai. > A question goes in; an adversarially reviewed report comes out, with cited > claims checked against their sources. Sources are retained in a > searchable workspace vault. A hosted remote MCP server and a standalone > citation-verification API are available. Runs are priced flat per run, in > two sizes: Light $9 and Deep $49. - API base: https://api.hyperresearch.ai/v1 - OpenAPI 3.1 document: https://api.hyperresearch.ai/v1/openapi.json - Remote MCP server (streamable HTTP): https://mcp.hyperresearch.ai/mcp - Console: https://console.hyperresearch.ai - Open-source CLI: https://github.com/jordan-gibbs/hyperresearch - Every page under /docs is also available as Markdown by appending `.md`, for example https://hyperresearch.ai/docs/quickstart.md - Sign-ups are open at https://console.hyperresearch.ai/login ## Benchmark On 10 tasks from DRACO, a deep research benchmark from Perplexity and Harvard, graded by us with the same judge (GPT-5.2 at the paper's settings) in October 2026, Hyperresearch Deep scored 77.7, the highest of the systems we tested; Claude Research (Opus) scored 73.0. This is our measurement, not an official leaderboard. Method, every task score and the limits: https://hyperresearch.ai/benchmarks/draco On 100 BrowseComp questions, asked through the production API between October 29 and 31, 2026 and graded by us with one grader (GPT-5.6 Luna) for every system, Hyperresearch Answers answered 89 correctly; Parallel Pro answered 73. On 100 DeepSearchQA questions it answered 85, against 81 for Parallel Pro. This is our measurement, not either benchmark's official protocol. Method, grader, one traced question and the limits: https://hyperresearch.ai/benchmarks/answers ## 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 `, 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). - 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. ## Start here - [Hyperresearch](https://hyperresearch.ai/): what the service does, the pipeline stages, and the four product surfaces. - [Pricing](https://hyperresearch.ai/pricing): Light $9, Deep $49, flat per run, pay as you go. A Light run takes about 30 to 40 minutes and keeps about 20 to 50 sources. - [Documentation overview](https://hyperresearch.ai/docs): the hosted API, start to finish. ## Documentation - [Quickstart](https://hyperresearch.ai/docs/quickstart): create a scoped key and submit a run. - [Authentication](https://hyperresearch.ai/docs/authentication): console sessions and bearer API keys. - [Runs](https://hyperresearch.ai/docs/runs): run sizes, idempotency, rerun-from, blocked runs. - [Vault](https://hyperresearch.ai/docs/vault): FTS, semantic and hybrid search, export and import. - [Verification](https://hyperresearch.ai/docs/verification): check a report's citations, including documents Hyperresearch did not write. - [MCP server](https://hyperresearch.ai/docs/mcp): endpoint, OAuth with PKCE, scopes. - [Webhooks](https://hyperresearch.ai/docs/webhooks): endpoints, signing, deduplication. - [Errors and retries](https://hyperresearch.ai/docs/errors): what is safe to retry and what each status means. - [Versioning and deprecation](https://hyperresearch.ai/docs/versioning): /v1 stability, Deprecation and Sunset headers, notice periods. - [Billing](https://hyperresearch.ai/docs/billing): usage, spend cap, invoices. - [API reference](https://hyperresearch.ai/docs/api): one page per operation, generated from the OpenAPI document. ## Product - [Answers MCP](https://hyperresearch.ai/answers): two MCP tools for agents at https://answers.hyperresearch.ai/mcp. search returns ranked raw sources from the open web, 120+ structured sources and licensed news articles with word-for-word passages; ask returns a short answer checked against quotes, or not found. A licensed article is read under the publisher's own licence, quoted and cited, and not stored. Pay per call from a prepaid wallet. The same calls over REST: POST https://api.hyperresearch.ai/v1/answers and POST /v1/answers/search, API key with the answers:ask scope ([reference](https://hyperresearch.ai/docs/api/operations/tags/answers)). Rate limit: 60 calls a minute per key and 120 per workspace (ask and search together); over it the API answers 429 with Retry-After, and a refused call is not charged. - [How it works](https://hyperresearch.ai/how-it-works): the pipeline a run goes through, step by step. - [Trust](https://hyperresearch.ai/trust): retention, workspace isolation, subprocessors. - [The vault](https://hyperresearch.ai/vault): sources accumulate across runs. - [Sources](https://hyperresearch.ai/sources): every source a production run can read, by subject: the web, papers and preprints, SEC filings and US stock prices, federal rules, bills, dockets, campaign finance and lobbying, health and life-science databases (ClinicalTrials.gov, openFDA, DailyMed, PubChem, UniProt), US and international statistics (Census, BLS, FRED, World Bank, IMF, OECD, Eurostat), weather and climate (NWS, NOAA, NASA), books and archives (Library of Congress, Internet Archive), code and package registries, patents (USPTO, EPO) and licensed news for research runs and Answers calls. Also at https://hyperresearch.ai/sources.md. - [The verification receipt](https://hyperresearch.ai/verification-receipt): every check a report passes before it ships. - [Citation verification API](https://hyperresearch.ai/citation-verification-api): verify documents Hyperresearch did not write. - [Open-source CLI](https://hyperresearch.ai/cli): the same pipeline on your own machine. ## Guides - [Research guides](https://hyperresearch.ai/guides): every guide, each built from public primary data with every figure linked to its source. Buying or sizing a small business: - [Commercial due diligence for a small acquisition](https://hyperresearch.ai/guides/commercial-due-diligence-small-acquisition): what commercial due diligence covers and costs, with a worked industry section for a $4M residential HVAC company from Census, BLS, EPA and SEC data. - [Is a laundromat a good investment?](https://hyperresearch.ai/guides/is-a-laundromat-a-good-investment): eleven claims checked against 25 years of Census and BLS data. - [Self-storage feasibility study](https://hyperresearch.ai/guides/self-storage-feasibility-study): what a feasibility study costs and contains, and a do-it-yourself demand check from Census data and REIT 10-Ks, worked for Tucson. - [TAM SAM SOM](https://hyperresearch.ai/guides/how-to-calculate-market-size): formulas, top-down vs bottom-up, and a bottom-up worked example from Census counts. - [SBA loan to buy a business](https://hyperresearch.ai/guides/sba-loan-to-buy-a-business): the SBA 7(a) acquisition rules under SOP 50 10 8.1 (equity, seller notes, guarantees, collateral, rates and fees) with a worked deal. - [SBA 7(a) acquisition loan data](https://hyperresearch.ai/guides/sba-7a-acquisition-loan-data): SBA loans to buy a business, FY2019 to FY2026, from SBA's loan-level file: size, rate spread, industries, lenders, states, charge-offs, CSV downloads. - [Business broker fees](https://hyperresearch.ai/guides/business-broker-fees): sourced commission ranges, Lehman-style scales, minimums, retainers and tails, with worked math. - [Buying a franchise](https://hyperresearch.ai/guides/buying-a-franchise): Item 7 costs from four current FDDs, royalty and ad fund math, SBA directory rules, territory, renewal and a dated checklist. Judging a market research report: - [How to check a market research report before you buy](https://hyperresearch.ai/guides/how-to-check-a-market-research-report): the checks to run on a syndicated report, with one headline figure traced to Census and SEC data. - [Research subscription costs and cheaper routes](https://hyperresearch.ai/alternatives/research-subscription-costs): what IBISWorld, Statista, Mintel, Gartner, PitchBook and others charge, read from their own pages and filings. - [What PitchBook costs](https://hyperresearch.ai/alternatives/pitchbook-pricing): PitchBook seat prices from US federal purchase orders and Morningstar's 10-K, Gartner's government price list, CB Insights orders, and the free routes. - [Is Statista reliable?](https://hyperresearch.ai/guides/is-statista-reliable): thirteen public Statista figures traced to their original sources. - [Why market size reports disagree](https://hyperresearch.ai/guides/why-market-size-reports-disagree): eight firms' estimates for one market, restated and checked against the Economic Census. Companies and industries from public filings: - [AI coding agents in SEC filings](https://hyperresearch.ai/guides/ai-coding-agents-in-sec-filings): which public companies name GitHub Copilot, Cursor, Claude Code and other AI coding tools in 10-K, 10-Q, 8-K and S-1 filings, per quarter since 2021, CSVs. - [Account research from a 10-K](https://hyperresearch.ai/guides/account-research-from-a-10-k): where the useful parts of a 10-K sit, and a worked account brief on a mid-cap insurer cited to section. - [Tariff impact on medical equipment](https://hyperresearch.ai/guides/tariff-impact-medical-equipment): which US tariffs apply to imported medical equipment, where import exposure sits by country, and what device makers report in their filings. Research tools and methods: - [Out of deep research?](https://hyperresearch.ai/alternatives/deep-research-limit-reached): how ChatGPT, Gemini and Claude meter deep research, and when one report beats upgrading. - [OpenAI deep research API alternatives](https://hyperresearch.ai/alternatives/openai-deep-research-api): what replaces the retired o3-deep-research and o4-mini-deep-research models, what each option costs, and how the calls map to /v1/runs. - [Perplexity deep research compared](https://hyperresearch.ai/alternatives/perplexity-deep-research): Perplexity Research mode and the Sonar Deep Research API (moved to the Agent API on 2026-09-27) against Hyperresearch on price, time, citations and API access. - [Gemini Deep Research compared](https://hyperresearch.ai/alternatives/gemini-deep-research): the Gemini app and the Gemini API deep research agent against Hyperresearch, with where each one is the better choice. - [Claude Research compared](https://hyperresearch.ai/alternatives/claude-research): Claude Research on paid Claude plans against Hyperresearch on plans, limits, API access and DRACO scores. - [Exa alternatives](https://hyperresearch.ai/alternatives/exa-alternatives): nine search APIs for AI agents (Exa, Tavily, Parallel, Linkup, Perplexity, Brave, You.com, Firecrawl, Hyperresearch Answers) on price per call, links or a cited answer, hosted MCP, free allowance and rate limits. - [How to add web search to an AI agent](https://hyperresearch.ai/guides/add-web-search-to-ai-agent): a hosted MCP server, a REST tool you define, or a ready-made AI SDK or LangChain tool, with working code and which shape to pick. - [Claude Code as a research harness](https://hyperresearch.ai/guides/claude-code-research-harness): the open-source CLI in Claude Code, and how skills, subagents and tool locks keep a long run on track. - [Legal open-access full text for agents](https://hyperresearch.ai/guides/open-access-full-text-for-agents): free APIs that return the full text of a paywalled DOI, or say there is none. - [Retraction-aware research](https://hyperresearch.ai/guides/retraction-aware-research): checking a DOI's retraction status for free through Crossref. - [Verify citations in AI output](https://hyperresearch.ai/guides/verify-citations-in-ai-output): the four ways citations in machine-written text go wrong, and how to catch each. - [Cost per verified citation](https://hyperresearch.ai/guides/cost-per-verified-citation): a unit for comparing research APIs. ## MCP clients - [Claude](https://hyperresearch.ai/mcp/claude), [Cursor](https://hyperresearch.ai/mcp/cursor), [ChatGPT](https://hyperresearch.ai/mcp/chatgpt) - [Long-running runs over MCP](https://hyperresearch.ai/mcp/long-running-runs) ## Docs digests - [/docs/llms.txt](https://hyperresearch.ai/docs/llms.txt): the documentation link map on its own. - [/docs/llms-full.txt](https://hyperresearch.ai/docs/llms-full.txt): every documentation page concatenated as Markdown. ## Optional - [About](https://hyperresearch.ai/about): what Hyperresearch is, who builds it, and the entity details. - [Contact](https://hyperresearch.ai/contact): support@hyperresearch.ai; enterprise inquiries through https://hyperresearch.ai/enterprise. - [Terms of Service](https://hyperresearch.ai/legal/terms) - [Privacy Policy](https://hyperresearch.ai/legal/privacy) - [Verify a report](https://hyperresearch.ai/verify): the public citation-verification demo, three submissions per network per day.