GPT-4V (1106 preview): benchmark results
Provider: OpenAI. Access: API.
Aggregate Rating 1578 ± 22, rank #1275 of 10916 rated models, from 153 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| HELM Image2Struct - Image2Latex Table | 72.34 | EMS | 100 |
| HELM VHELM - Blink Perception - Relative Reflectance | 68.66 | Prefix Quasi-Exact Match (%) | 95.7 |
| HELM VHELM - Bias | 89.36 | Mean win rate | 89.4 |
| HELM VHELM - Mmmu - Art Theory | 93.33 | Prefix Quasi-Exact Match (%) | 89.4 |
| HELM VHELM - Pairs | 91.58 | PEM | 89.4 |
| HELM VHELM - Crossmodal 3600 - Spanish Locale | 3.67 | Prometheus Vision Rating (1-5) | 87.2 |
| HELM VHELM - Blink Reasoning - IQ Test | 48.67 | Prefix Quasi-Exact Match (%) | 81.9 |
| HELM VHELM - Exams V - Spanish (Natural Science) | 76.92 | Prefix Quasi-Exact Match (%) | 80.9 |
| HELM VHELM - Crossmodal 3600 - Swahili Locale | 3.37 | Prometheus Vision Rating (1-5) | 77.7 |
| HELM VHELM - Bingo - T2I | 3.42 | Prometheus Vision Rating (1-5) | 75.5 |
| HELM VHELM - Blink Reasoning | 77.11 | PEM | 75.5 |
| HELM VHELM - Mme - Posters | 95.92 | Prefix Quasi-Exact Match (%) | 74.5 |
Interactive version: theaggregate.ai/model?slug=gpt-4v-1106-preview · How It Works · Data refreshed daily, snapshot 2026-11-08.