Written and maintained by CASRAI Editorial Board
Last updated
Who this is for: researchers, librarians and research-data staff deciding whether GDELT suits a media or event-analysis project, and how to reach its data. Facts below come from the GDELT Project’s own pages, read on 2026-11-07.
What does GDELT stand for?
GDELT stands for Global Database of Events, Language and Tone. The GDELT Project describes it as an open database of human society, built from global news media, with archives spanning more than 215 years. Searches also use the name “GDELT Project”, which is the same thing: the project is the initiative, GDELT is the data it publishes.
Who runs it and what it covers
The project’s About page names Kalev H. Leetaru as its creator. Its data page lists several datasets rather than one table:
- Events: coded records of events reported in news media.
- Global Knowledge Graph (GKG): links people, organisations, locations, themes and emotions found in coverage.
- Mentions (added in GDELT 2.0): ties each event to the articles that mention it.
- Other graphs and indexes: Visual GKG, Frontpage Graph, Difference Graph, Entity Graph, Quotation Graph, Geographic Graph and TV Ngrams.
The GDELT 2.0 announcement (19 February 2015) says live translation covers 65 languages, 98.4% of non-English monitoring volume, and that the GCAM emotion and theme system has 24 packages and more than 2,300 emotions and themes. Update frequency differs by dataset: GDELT 2.0 Events and GKG every 15 minutes, GDELT 1.0 daily, Frontpage Graph hourly, TV Ngrams daily.
How to access GDELT (APIs, BigQuery, files)
The data page lists five routes. Pick by how much programming you want to do:
| Route | What the data page says it is | Version or scope |
|---|---|---|
| Raw CSV files | Downloadable files; the 2015 GKG alone exceeds 2.5 TB | GDELT 1.0 and 2.0 |
| Google BigQuery | Standard SQL against live tables updated every 15 minutes | GDELT 2.0 Event, Mentions and GKG tables |
| DOC, GEO and TV APIs | Return live JSON | News text, geographic and television data |
| GDELT Summary | A non-technical web interface built on those APIs | Browser use, no code |
| GDELT Analysis Service | Free browser visualisation and export | Searches GDELT 1.0 only, per the page |
How to use GDELT?
For a first look, use GDELT Summary or the Analysis Service from the project’s data page. For repeatable analysis, query BigQuery with SQL, or call the JSON APIs from your own scripts. Use the raw files only if you need to hold the data locally and have the storage for it. Query syntax for each interface is in the project’s own documentation, linked from the data page.
Is the GDELT API free?
Yes, as the project states it. The data page calls GDELT “100% free and open”, and the About page says datasets may be used for any purpose, academic, commercial or governmental. Redistribution is allowed, but you must cite the GDELT Project and link to https://www.gdeltproject.org/. The About page states no rate limits. BigQuery is a Google service, so its own usage terms and billing apply to your queries separately from GDELT’s licence.
Is GDELT still active?
The GDELT Project’s data page, read 2026-11-07, describes GDELT 2.0 Events and GKG as updating every 15 minutes. A separate listing in the Registry of Open Data on AWS describes its copy as “Not currently being updated” and “Unmanaged”. That flag applies to the AWS-hosted copy as listed in the registry; it is not a statement from gdeltproject.org about the live feeds. If your work depends on freshness, take the data from the project’s own routes (BigQuery, APIs or files) rather than the AWS copy, and check the latest timestamps before relying on them.
Strengths and limits
- Strength: scale and frequency. Machine-coded global news coverage refreshed every 15 minutes in GDELT 2.0, with translation across 65 languages.
- Strength: open terms. Free for any purpose with attribution; see our entry on open data for the general concept.
- Limit: it measures news coverage, not events directly. Records are derived from media reports, so they reflect what outlets covered and how. That is an editorial interpretation researchers should account for in their methods section.
- Limit: version differences. The Analysis Service searches only GDELT 1.0, while BigQuery carries the 2.0 tables; check which version a tool queries before comparing results.
- Limit: size. Raw files are large, so BigQuery or the APIs are usually the practical route.
Citing and documenting GDELT in research
Record the dataset (Events, GKG or Mentions), version (1.0 or 2.0), access route and retrieval date, and include the required GDELT Project citation and link. Good metadata for the extract makes the analysis reproducible; where you share derived files, see data sharing agreement for governing terms. For another open incident-style data source, see our comparison of OECD AIM and the AI Incident Database.
FAQ
What does GDELT stand for?
Global Database of Events, Language and Tone.
Is GDELT still active?
The project’s data page describes 15-minute updates for GDELT 2.0; the AWS registry copy is flagged as not currently updated. See the section above.
Is the GDELT API free?
Yes. The project describes its data as free and open for any purpose, with a required citation and link.
Is GDELT the same as the GDELT Project?
In practice yes: the GDELT Project is the initiative that publishes the GDELT datasets.
Sources
- GDELT Project, Data: gdeltproject.org/data.html (read 2026-11-07)
- GDELT Project, About: gdeltproject.org/about.html (read 2026-11-07)
- GDELT 2.0: Our Global World in Realtime: blog.gdeltproject.org (read 2026-11-07)
- Registry of Open Data on AWS, GDELT: registry.opendata.aws/gdelt (read 2026-11-07)
