AI Release RadarRadar prioritizes pages controlled by the releasing lab: model pages, docs, changelogs, newsrooms and official blogs.
About the Radar
A source-first index of model launches.
AI Release Radar tracks named model releases from major AI labs and explains what changed, where the evidence came from and what still needs review.
Editorial principles
How entries earn a place.
Radar is intentionally narrow: it indexes named model releases, not every AI product update.
Release notes are normalized into the same fields: capability, availability, pricing, context and evidence confidence.
If a claim is not clearly present in the source, the field stays unknown or the release is marked Needs Review.
Included
What counts.
- Named public model launches or model-family updates.
- Public API, app, open-weight, research or preview releases from a major AI lab.
- Updates with a public announcement or availability date.
- Existing entries with clearer new first-party documentation.
Excluded
What stays out.
- Rumours, leaks, private chatter or benchmark-only claims.
- Silent aliases, infrastructure-only products or vague capability marketing.
- Minor product UI changes with no named model release.
- Third-party summaries unless they lead back to a first-party source.
Reading the labels
How to interpret the Radar.
Each release page separates evidence strength from practical availability and comparable change notes.
Flags whether a release is Major, Meaningful, Incremental or Needs Review based on source strength and practical change.
Explains how strong the evidence is: model card, changelog, launch announcement, social-only source or Needs Review.
Shows what users can practically do with the model: status, channel, weights, pricing and docs.
Turns each launch into comparable fields so releases from different labs can be read side by side.
Maintenance
Several checks daily,
human caution.
Radar checks monitored first-party sources several times daily and updates the index when a release qualifies or when clearer information improves an existing Needs Review entry.
It is maintained by Mahdi Taghizadeh. If a model is missing, the most likely reason is that the available source has not yet provided enough comparable information.