LLaVA-Med-v1.5-7B: benchmark results
Provider: LLaVA Team.
Aggregate Rating 1485 ± 49, rank #2593 of 10916 rated models, from 12 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Holtercare-Bench - Diagnosis (Text) | 27.95 | Accuracy (%; multiple-choice closed QA on the overall diagno | 41.7 |
| MedCTA (No Tools) | 39.25 | Goal accuracy (%; final answer judged against the clinician- | 38.9 |
| MMBU - Grounded Classification (Boxes) | 34 | Closed-ended classification of a region marked by a bounding | 31.2 |
| Holtercare-Bench - Event Counting (Text) | 22.32 | Accuracy (%; multiple-choice closed QA on how many times a g | 25 |
| MMBU - Grounded Classification (Boxes, Open-Ended) | 3.8 | Open-ended classification of a region marked by a bounding b | 25 |
| MMBU - Grounded Classification (Masks, Open-Ended) | 3.5 | Open-ended classification of a region marked by a segmentati | 25 |
| MMBU - Ungrounded Classification | 39.5 | Closed-ended (multiple-choice) ungrounded classification of | 25 |
| MMBU - Ungrounded Classification (Open-Ended) | 3.3 | Open-ended (free-form) ungrounded classification of the whol | 25 |
| Holtercare-Bench - Event Timing (Text) | 25.91 | Accuracy (%; multiple-choice closed QA on the millisecond ti | 16.7 |
| Holtercare-Bench - Presence (Text) | 43.79 | Accuracy (%; multiple-choice closed QA on whether a given rh | 8.3 |
| Holtercare-Bench - Heart Rate Extremum Timing (Text) | 12.1 | Accuracy (%; multiple-choice closed QA on the millisecond ti | 0 |
| MMBU - Grounded Classification (Masks) | 8 | Closed-ended classification of a region marked by a segmenta | 0 |
Interactive version: theaggregate.ai/model?slug=llava-med-v1-5-7b · How It Works · Data refreshed daily, snapshot 2026-11-08.