Diffbot
Diffbot uses AI models to turn unstructured web content into linked knowledge.
What Diffbot is
Diffbot uses AI models to turn unstructured web content into linked knowledge. Extract classifies pages and returns structured JSON; Crawl starts from seed URLs and collects extracted pages in structured datasets. Natural Language resolves entities, relationships, coreferences, salience, and sentiment in text, while the Knowledge Graph supports structured queries about companies, people, news, and other public web entities. Web Search offers an API and a self-hostable search index designed for a machine or cloud, including offline use. Its page says closed media sites such as Facebook, Reddit, LinkedIn, and Instagram are generally off limits. Diffbot lists Python and TypeScript libraries and agent skills as interfaces, and supports API, self-hosted, and web platforms. The Free plan includes all four APIs, 10,000 credits per month, and up to five requests per minute, without a credit card. It returns a 429 Quota Exceeded response after the credit allowance or rate limit is reached; quotas reset at the start of each billing period.
Good
- Free plan includes all four APIs.
- Extract returns structured JSON from classified pages.
- Crawl collects extracted pages into structured datasets.
- Web Search supports self-hosting and offline use.
Watch out
- Free plan is limited to 10,000 credits per month.
- Free plan allows up to five requests per minute.
- Closed media sites are generally off limits to Web Search.
Who it suits
Diffbot suits companies seeking to access unstructured public web data at scale. Developers can use its APIs, listed language libraries, and agent skills.
Verdict
Diffbot offers several ways to extract, query, and search public web information, including a free tier. Check the free tier’s credit and request limits, and note the stated restrictions on closed media sites.
Plans and prices
From Diffbot's own pricing page, checked 7 Oct 2026.
| Plan | Price | What it gives you |
|---|---|---|
| Free | Free | All four APIs included · 5 requests / min · No credit card required · 10,000 credits / month |
| Startup | $299 a month | 250,000 credits / month · Overage at $0.001 / credit · All four APIs included · 5 requests / sec · $299 / mo |
| Plus | $899 a month | 1,000,000 credits / month · Overage at $0.0009 / credit · Access to Crawl (25 active) · 3 user seats · 25 requests / sec · $899 / mo |
| Enterprise | Priced on request | Custom credit allotment · 100+ active crawls · Custom user seats · Dedicated support · 25+ requests / sec · Custom |
The facts
Each one read from Diffbot's own pages, with the date it was checked.
What it does
- Crawl
- Crawl spiders sites from seed URLs and extracts pages into structured datasets.checked 7 Oct 2026
- Extract
- Extract classifies a page and returns structured JSON based on a standard page type and ontology.checked 7 Oct 2026
- Integrations
- The documentation lists Python and TypeScript libraries and agent skills as interfaces.checked 7 Oct 2026
- Knowledge Graph
- The Knowledge Graph supports structured queries about companies, people, news, and other public web entities.checked 7 Oct 2026
- Natural Language
- The Natural Language service resolves entities, relationships, coreferences, salience, and sentiment in plain text.checked 7 Oct 2026
- Security and privacy
- Diffbot’s privacy policy says it issues subscribers a unique API key and collects login, query history, and API call data to provide the services.checked 7 Oct 2026
- Support
- Technical and implementation help is available at [email protected], while account, billing, and sales inquiries go to [email protected] or the listed phone number.checked 7 Oct 2026
- Target users
- Diffbot says its services are used by companies worldwide to access unstructured public web data at scale.checked 7 Oct 2026
- Web Search
- Diffbot Web Search offers an API and self-hostable search index designed to run on a machine or cloud, including offline use.checked 7 Oct 2026
- Web Search limit
- The Web Search page says closed media sites such as Facebook, Reddit, LinkedIn, and Instagram are generally off limits.checked 7 Oct 2026
- What it does
- Diffbot builds AI models that read and transform the unstructured web into linked knowledge.checked 7 Oct 2026
The company
- Headquarters
- The company lists its address as 333 Ravenswood Ave, Menlo Park, CA 94025.checked 7 Oct 2026
Money
- Plan limits
- The Free plan returns a 429 Quota Exceeded response after 10,000 credits or its rate limit, and quotas reset at the start of each billing period.checked 7 Oct 2026
Alternatives to Diffbot
- 1
Amazon Textract AnalyzeExpense
Free planWebAPISelf-hosted7.6From $0.01 a month - 2
Weka
Free planOpen sourceAPILinuxmacOSSelf-hosted7.5Free plan - 3
Veryfi API
Free planAndroidiOSWebAPI7.4From $500 a month - 4
KNIME Analytics Platform
Free planLinuxmacOSWindows7.1From $19 a month - 5
Nanonets Approval Routing Agent
Free planWebAPISelf-hosted6.9Free plan - 6
Octoparse
Free planAPImacOSWindows6.8From $69 a month