URSA-8B — benchmark results
Provider: Other. Access: Open.
Unified ELO 1448 ± 13, rank #1056 of 1841 rated models, from 99 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| OpenVLM MathVista - GPS | 81.7 | Accuracy (%) | 91.7 |
| OpenVLM MathVista - GEO | 79.5 | Accuracy (%) | 91.3 |
| OpenVLM MathVista - ALG | 73.7 | Accuracy (%) | 84.4 |
| Open LMM Reasoning - MathVision - combinatorial geometry | 26.6 | Accuracy (%) | 84.2 |
| Open LMM Reasoning - DynaMath - Subject-plane geometry | 31.2 | Accuracy (%) | 82.1 |
| Open LMM Reasoning - MathVision - metric geometry - length | 36.1 | Accuracy (%) | 78.6 |
| OpenVLM MathVista - MWP | 75.3 | Accuracy (%) | 74.8 |
| Open LMM Reasoning - WeMath - Understanding of Plane Figures | 65.8 | Accuracy (%) | 74.1 |
| Open LMM Reasoning - MathVision - metric geometry - angle | 35.3 | Accuracy (%) | 74 |
| Open LMM Reasoning - MathVision - transformation geometry | 27.4 | Accuracy (%) | 74 |
| Open LMM Reasoning - MathVision - metric geometry - area | 33.6 | Accuracy (%) | 73.5 |
| Open LMM Reasoning - MathVision - solid geometry | 26.2 | Accuracy (%) | 73 |
Interactive version: theaggregate.ai/model?slug=ursa-8b · How It Works · Data refreshed daily, snapshot 2026-07-25.