APITube News API
The News API returns news articles from 500,000+ sources worldwide, already enriched with named entities, sentiment, IPTC categories, topics, industries and readability scores. Everything is served over HTTPS from https://api.apitube.io and needs an API key.
This page is the map of the section. If you already know what you are looking for, the full endpoint list lives in Endpoints.
Your first request
curl -H "X-API-Key: YOUR_API_KEY" \
"https://api.apitube.io/v1/news/everything?title=bitcoin&per_page=3"Get a key at apitube.io, then read Authentication for the three ways to pass it (X-API-Key header, Authorization: Bearer, or the api_key query parameter).
Articles come back under results, and each article links to its page through href — the full object is described in API Response Structure.
Sections of this reference
| Section | What is in it |
|---|---|
| Getting started | The endpoint list, authentication, response structure, rate limits, points and error codes. |
| Articles | Every endpoint that returns articles — search, top headlines, browse by taxonomy, local, raw, by ID, story clusters. |
| Insights | Answers about the corpus instead of a feed: trends, counts, fact-check verdicts. |
| Directories | Profiles of the people, companies and journalists behind the news, with coverage stats. |
| Real-time delivery | SSE, WebSocket and webhooks — get articles as they are indexed instead of polling. |
| Reference data | Allowed filter values: countries, languages, categories, topics, industries, event types, autocomplete. |
| AI | Natural language search, MCP server, code editor setup, agent skills. |
| Tools | Query builder, API explorer, Postman collection, SDKs and integrations. |
| Help | Glossary, technical FAQ, debugging, status page. |
How the API is shaped
Three things explain most of the surface:
One endpoint does the searching. /v1/news/everything accepts the complete filter set. Every other article endpoint is that same search with one value moved into the path or one restriction applied — /v1/news/category/{taxonomy}/{category_id} fixes the category, /v1/news/top-headlines fixes a source-quality floor. Learn the filters once and they work everywhere.
Filtering is on resolved values, not strings. People, organizations, brands and places are recognised during enrichment and get IDs, so organization.name=Tesla is not a text match. When you need an ID, the autocomplete endpoints resolve a prefix to one.
Requests are priced in points. A normal request costs 1 point; the prompt parameter adds 2 for the translation step, and fact-check costs 5. /v1/balance reports what is left and never costs anything itself. Points are separate from the per-minute rate limit.
Next steps
- Pick an endpoint: Endpoints
- See the filters in action: Usage examples
- Build a request in the browser: Interactive Query Builder
- Coming from another provider: Migrations