Diffbot

Diffbot is a set of web data APIs that gives AI applications structured knowledge from the public web instead of raw markdown. Its tools cover page extraction, site crawling, natural language processing, a Knowledge Graph of companies, people and news, and a web search API. The Extract API classifies a single URL and returns structured JSON, and the vendor states it is not an LLM. Crawl walks a whole site from one seed URL and collects thousands of pages into a dataset. The natural language API resolves entities, relationships, facts and sentiment from plain text, and the Knowledge Graph accepts structured queries. The search API is described as ad free and free of telemetry, and it can be self hosted from a packaged 4TB index or used through the hosted API. The vendor reports P90 latency under 300 milliseconds for hosted queries. It is aimed at developers and teams grounding agents and applications in verified web facts, with sample workflows such as sales briefs, financial due diligence and OSINT. Pricing is credit based. Free costs $0 per month with 10,000 credits, all four APIs and 5 requests per minute, with no credit card required. Startup costs $299 per month with 250,000 credits, overage at $0.001 per credit and 5 requests per second. Plus costs $899 per month with 1,000,000 credits, overage at $0.0009 per credit, access to Crawl for 25 active crawls, 3 user seats and 25 requests per second. Enterprise is custom. Extracting a page costs 1 credit, a web search 1 credit and exporting a Knowledge Graph entity 25 credits.