Deep research API

Valyu DeepResearch in Best Deep Research AI

Answer in brief

Valyu DeepResearch is an async API that runs multi-step research over the open web and proprietary sources, returning a cited report plus optional PDF, DOCX, XLSX and PPTX files.

Best for

Developers who want cited reports, office-file deliverables and per-task pricing

Pricing

Each task reserves its base credit price when created: instant 8, fast 25, standard 100, heavy 500 and max 2,500. Tool add-ons such as screenshots, code runs and browser sessions bill on top, and failed tasks are not charged. Cancelling a running task still charges the full base price, and Valyu's launch post lists CPM and per-request prices that do not match the credit schedule in the API docs.

What it covers

The open web plus domain sources such as financial filings, biomedical literature, patents, economic indicators, legal records and academic publishing. Users can restrict sources, upload their own files and connect MCP servers.

Citations and verification

Every claim links to the retrieved passage it came from. Valyu says numerical claims are verified before they enter the deliverable, and webpage screenshots serve as visual citations.

API

POST /v1/deepresearch/tasks starts an async task that returns a deepresearch_id; poll the status endpoint or set a webhook_url signed with HMAC-SHA256. Default limits are 10 requests per second per organization and 86,400 per day (https://docs.valyu.ai/api-reference/endpoint/deepresearch-create.md).

Strengths and limitations

Strengths

Fixed, mode-based task pricing with a cost breakdown on completed tasks
Report plus PDF, DOCX, XLSX, PPTX and structured JSON from one call
Human checkpoints at planning, retrieval, synthesis and deliverable stages

Limitations

Credits have no dollar value in the docs, so cost per task needs a conversion
Most benchmark numbers are Valyu's own runs, and they conflict across sources
Max mode can run up to three hours, per Valyu's launch post

Third-party ratings

Not found

Dated facts

Valyu's launch post gives no calendar date; the create docs were read on 2026-11-08.
Vendor-reported: 72.7 on DRACO in Heavy mode. Hyperresearch's October 2026 DRACO run scored Heavy at 57.2 on 8 of 10 tasks.
Vendor-reported: 74.5 on ScholarQA-CS, 29 percent above OpenScholar, per Valyu's ScholarQA post.

Valyu DeepResearch or Hyperresearch?

Choose Valyu DeepResearch for developers who want cited reports, office-file deliverables and per-task pricing. For sourced research where every figure has to hold up, Hyperresearch is our editorial choice: it scored 77.7 on DRACO in our own run and checks each cited figure against its source.

Sources for this entry

Vendor descriptions are attributed to the vendor. Prices change; confirm with each vendor.

Continue this reading path

Cite this entry

Best Deep Research AI. “Valyu DeepResearch in Best Deep Research AI.” Hyperresearch reference, reviewed 8 November 2026.
https://bestdeepresearchai.com/resources/valyu-deepresearch/