LLaVA-OneVision-7B: benchmark results
Provider: LLaVA Team. Released 2024-08-05. Access: Open.
Aggregate Rating 1479 ± 10, rank #2674 of 10916 rated models, from 1360 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| DeepTumorVQA - Medical Reasoning | 39.4 | Accuracy (%) on the Medical-reasoning-stage questions (two t | 100 |
| FGTR-Bench - Road Participants | 53.3 | Accuracy (%) on 1,200 road-participant questions, 4-option m | 100 |
| OpenVLM MMBench V1.1 CN - Image Emotion | 90 | Accuracy (%) | 100 |
| RealStreamEval - Backward Tracing | 51.5 | Score (%; RealStreamEval on OVO-Bench tasks and questions: f | 100 |
| RepProbe Synthetic - OBO Mid | 13.3 | OBO Mid (%) | 100 |
| OpenVLM MMT-Bench - AUD | 70 | Score (%) | 99.5 |
| OpenVLM MMT-Bench - Scene Graph Recognition | 100 | Score (%) | 99.3 |
| MEGA-Bench Task - Red Teaming Visual Misleading | 91.1 | Task Score (%) | 98.8 |
| OpenVLM MMBench V1.1 EN - Attribute Recognition | 97.8 | Accuracy (%) | 98.1 |
| OpenVLM MMT-Bench - Multiple View Image Understanding | 50 | Score (%) | 98.1 |
| OpenVLM MMT-Bench - Plant Recognition | 95 | Score (%) | 98.1 |
| OpenVLM MMT-Bench - Handwritten Mathematical Expression Recognition | 100 | Score (%) | 97.8 |
Interactive version: theaggregate.ai/model?slug=llava-onevision-7b · How It Works · Data refreshed daily, snapshot 2026-11-08.