- What changed
- RADAR, an expert-level generalist vision-language model for abdominal CT diagnosis trained on over 400,000 exams, has been released with code and checkpoints.
- Why you should care
- Vision-language models trained on large volumes of unannotated medical imaging can achieve expert-level performance across clinical tasks.
- Your move
- Watch. Monitor independent evaluations of medical vision-language models for clinical reliability.
- What to watch next
- Independent clinical validation studies assessing RADAR performance on diverse hospital datasets.
- Event
- research
- Event date
- Sep 19, 2026
- Relevant to
- General AI readers