Keye-VL-8B: benchmark results
Provider: Other.
Aggregate Rating 1561 ± 34, rank #1526 of 10916 rated models, from 17 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| HID-Chart - Computer Science | 44.9 | Accuracy (%; GPT-5 mini judge, 323 computer science question | 60 |
| MathVision | 46 | Overall Accuracy (%) | 59.6 |
| HID-Chart - Finance | 42.5 | Accuracy (%; GPT-5 mini judge, 327 finance questions) | 53.3 |
| HID-Chart - Social | 38.8 | Accuracy (%; GPT-5 mini judge, 335 social questions) | 46.7 |
| HID-Chart | 40 | Accuracy (%; GPT-5 mini judge against the annotated answer, | 39.5 |
| HID-Chart - Chemistry | 46.3 | Accuracy (%; GPT-5 mini judge, 108 chemistry questions) | 36.7 |
| HID-Chart - Engineering | 37.9 | Accuracy (%; GPT-5 mini judge, 116 engineering questions) | 33.3 |
| DementiaCare-Bench - 128 Frames | 74.7 | Accuracy (%) | 27.3 |
| DementiaCare-Bench - 16 Frames | 75 | Accuracy (%) | 27.3 |
| DementiaCare-Bench - 32 Frames | 75 | Accuracy (%) | 27.3 |
| DementiaCare-Bench - 48 Frames | 74.8 | Accuracy (%) | 27.3 |
| DementiaCare-Bench - 64 Frames | 74.7 | Accuracy (%) | 27.3 |
Interactive version: theaggregate.ai/model?slug=keye-vl-8b · How It Works · Data refreshed daily, snapshot 2026-11-08.