Event Registry API for company news
Event Registry, sold as newsapi.ai, is a news index that resolves companies to Wikipedia concepts and groups articles into events. It has not been run on the company-news benchmark. This page covers how its concept search works, what tokens cost, and where it stands against the search APIs that have been measured.
The benchmark's 300 questions each name a company and ask for a recent fact about it. The measured APIs get that question as text. Event Registry is built to take an identifier instead, which is a different contract and the reason it belongs on this page rather than in the ranking.
Benchmark last measured . 300 questions.
Concepts, not keywords
The article endpoint is GET /api/v1/article/getArticles. It accepts a plain keyword, but the filter the product is built around is conceptUri. Event Registry runs an annotator over every article and links the people, places and organizations it finds to Wikipedia pages. The company you care about is therefore one URI, and an article in German that calls it by a translated name resolves to the same URI as a wire story in English. The /annotate endpoint exposes the same step for your own text, which is how you look a company up before you search for it.
Searches combine conceptUri, sourceUri, locationUri and date filters. A trial key allows 15 keywords and 15 concepts per query; paid plans allow 60 of each. At most five requests can be in flight at once on any plan.
Events are the second layer. GET /api/v1/event/getEventstakes the same concept filters and returns clusters: articles from different outlets that the system judged to be about one happening, with the cluster's own concepts, date and size. For a monitoring feed this is the useful unit. A funding announcement that thirty outlets cover arrives once.
Tokens, plans and what a search costs
- Free. 2,000 tokens, given once, no expiry. Last 30 days of data.
- 5K plan. $90 a month for 5,000 tokens, $18 per 1,000 tokens. Archive since 2014 starts here.
- 10K plan. $150 a month for 10,000 tokens, $15 per 1,000 tokens. A 20K plan sits above it.
- Per call. Article search over the last 30 days: 1 token. Archive article search: 5 tokens per year searched. Event search: 5 tokens recent, 20 per archive year. Aggregates and clustering cost 5 to 50 tokens.
- Overage. Calls are rejected when a plan runs out unless you allow extra usage at $0.015 a token.
For comparison with the measured APIs, the 5K plan works out to $18 per 1,000 recent searches if every call is a one-token article search. The measured web search APIs list between $0 and $16 per 1,000 queries. The difference is what you get back: Event Registry returns entity-linked articles with event membership, the search APIs return pages and snippets.
Vendor prices, limits and parameter names were read from the linked pages on 18 November 2026. They are the vendor's list terms, not measurements.
Where it sits against the measured APIs
The benchmark has not scored Event Registry, so there is no row for it. What the board does show is how far plain-text search gets on the same job. The top of the web search band is below; the full table is on the benchmark page.
Top five measured web search configurations on company-news questions
| Rank | Provider | Accuracy | Answer recall @5 | Mean search time | $ / 1k queries | $ / 1k correct |
|---|---|---|---|---|---|---|
| 1 | Exa (type=fast)POST /search · type=fast | 99.3% | 99.3% | 652ms | $7.00 | $7.05 |
| 2 | ContextPOST api.context.dev/v1/web/search · numResults=10 · highlights maxCharacters=1200 | 98.0% | 98.3% | 15.93s | $4.40 | $4.49 |
| 3 | Exa (type=instant)POST /search · type=instant | 97.7% | 97.3% | 398ms | $7.00 | $7.17 |
| 4 | PerplexityPOST /search · search_context_size=low | 97.3% | 98.3% | 1.38s | $5.00 | $5.14 |
| 5 | Linkup (depth=fast)POST /v1/search · depth=fast · outputType=searchResults | 96.7% | 94.7% | 1.57s | $5.00 | $5.17 |
Exa (type=fast) answers 99.3% of the questions with no identifier, no concept and no index of its own. That is the bar an entity-filtered news API has to clear on lookup. Where Event Registry should pull ahead is the part the lookup test does not measure: finding every article about the company over a month, in several languages, without namesakes, and collapsing them into events.
Running Event Registry on the benchmark means mapping each question's company to a concept URI first, then searching. That is one annotate call plus one search, or two tokens a question, and it fits inside the free allowance. It is on the list.
All news APIs and search APIs in one table →
Questions answered on this page
Has Event Registry been benchmarked?
No. Event Registry is not on the company-news board. The free plan's 2,000 tokens cover 2,000 searches over the last 30 days, which is enough to run the 300 questions once, so it is on the list of arms to add. Until then this page describes the API and makes no accuracy claim.
What does Event Registry cost?
Registration is free and gives 2,000 tokens once, with no expiry. The 5K plan is $90 a month for 5,000 tokens and the 10K plan is $150 a month for 10,000. A search over the last 30 days costs one token; searching the archive costs five tokens per year searched; an event search costs five tokens. Extra tokens on a paid plan are $0.015 each. Archive access back to 2014 starts at the 5K plan.
How is a concept different from a keyword?
A keyword matches text. A concept is a Wikipedia URI, so the company is one identifier regardless of how an article names it or which language it is written in. Searching conceptUri for a company returns articles the annotator linked to that entity and leaves out articles that only contain the string.
What is an event in Event Registry?
A cluster of articles from different sources about the same happening. The getEvents endpoint returns the cluster with its concepts, date and article count rather than the individual stories, which is what a monitoring workflow usually wants: one row per funding round, not thirty.
When would a web search API be the better choice?
For a single factual question about a named company. On the benchmark, Exa (type=fast) found the answer in its top ten results 99.3% of the time with the question sent as plain text and no entity identifier at all. Event Registry's advantage is on the monitoring side: a stable identifier, clustering, and archive depth, none of which the lookup benchmark scores.