Sheet music understanding on SSMR-Bench (test)
92.7AccuracyGemini 3.1 Pro
Evaluation Results
| Method | Links | |
|---|---|---|
| Gemini 3.1 ProContext=Legato 22026.07 | 92.7 | |
| Gemini 3.1 ProContext=Legato 12026.07 | 84.8 | |
| GPT-5Context=Legato 22026.07 | 71.7 | |
| Gemini 3.1 ProContext=None2026.07 | 71.4 | |
| GPT-5Context=Legato 12026.07 | 65.3 | |
| GPT-5Context=None2026.07 | 51.8 |