Physics Reasoning on QFT Synthetic Medium
88.8AccuracyGemini-2.5-flash
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Gemini-2.5-flashModel Type=Proprietary2026.04 | 88.8 | 97.5 | |
| Claude-Opus-4.5Model Type=Proprietary2026.04 | 85.2 | 93.8 | |
| OSS-20bModel Type=Open-Weight2026.04 | 76 | 93.8 | |
| Qwen3-4B-Thinking-2507Model Type=Open-Weight2026.04 | 49.5 | 68.8 | |
| Qwen3-4B-Instruct-2507Model Type=Open-Weight2026.04 | 33.2 | 46.2 | |
| DeepSeek-R1-Distill-Qwen-7BModel Type=Open-Weight2026.04 | 26.2 | 50 |