Answers by Context.dev

Answers is a research endpoint added to Context.dev's web-data API and announced by the company on 16 September 2026; its changelog lists it on 12 September, and it reached a wider audience through a launch post on the 20th. A request carries a task in plain language of up to 2,000 characters — naming a domain makes the research start there — and an example JSON object whose keys and placeholder values describe the shape wanted back. The service searches the web, reads the relevant pages and returns the answer in that shape as json_content, with the contributing URLs in sources in the order they were first seen, a partial flag if the deadline ended the research early, and the credits consumed. That structure is the point for a developer: the output can become fields in a record or rows in a comparison table without parsing prose, and the sources travel with it. There are two modes: fast, with a 30-second budget, at 10 credits, and ultra, with 50 seconds, at 100 credits; the API defaults to ultra and only successful requests are charged. The published plans make the arithmetic checkable: a free tier of 1,000 credits a month, Developer at $25 a month for 10,000 credits with overage at $15 per 10,000, Pro at $149 for 200,000 and Scale at $499 for 1,000,000 — so an ultra answer costs 100 of the Developer plan's 10,000 monthly credits. Zero data retention is available on the endpoint for organisations with that entitlement, and the company's changelog records SOC 2 Type II compliance on 18 September. The quality of an answer depends on what the pages it reads say, which the returned sources let a user check; no accuracy figure is published and none is recorded here. This is a capability of Context.dev, a web scraping and enrichment API, rather than a separate product.