Medical Report Generation on SpineBench
4.55SIPGemini-2.5-Pro
Evaluation results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| Gemini-2.5-ProSize=-2025.10 | 4.55 | 4.51 | 4.68 | 4.79 | 4.8 | 93.32 | 89.23 | |
| GPT5-miniSize=-2025.10 | 4.55 | 4.48 | 4.6 | 4.98 | 4.78 | 93.56 | 87.01 | |
| GPT5Size=-2025.10 | 4.54 | 4.51 | 4.53 | 4.69 | 4.64 | 91.6 | 85.54 | |
| Gemini-2.5-FlashSize=-2025.10 | 4.43 | 4.29 | 4.57 | 4.88 | 4.75 | 91.68 | 83.93 | |
| O3Size=-2025.10 | 4.39 | 4.25 | 4.34 | 4.43 | 4.42 | 87.32 | 85.36 | |
| SpineGPTSize=7B2025.10 | 4.15 | 4.1 | 4.41 | 4.54 | 4.62 | 87.24 | 87.44 | |
| Claude4Size=-2025.10 | 3.96 | 4.08 | 4.04 | 4.44 | 4.41 | 83.72 | 80.28 | |
| GLM-4.5VSize=21B2025.10 | 3.85 | 3.78 | 3.83 | 4.05 | 4.3 | 79.24 | 83.26 | |
| GPT-4oSize=-2025.10 | 3.16 | 3.03 | 3.06 | 3.3 | 3.46 | 64.04 | 81.6 | |
| Qwen2.5-VL-72BSize=72B2025.10 | 3.14 | 3.03 | 3.12 | 3.23 | 3.42 | 63.8 | 79.88 | |
| Linshu-32BSize=32B2025.10 | 3.05 | 3.05 | 3.23 | 3.49 | 3.47 | 65.16 | 77.3 | |
| Medgemma-27BSize=27B2025.10 | 2.88 | 3.49 | 3.38 | 4.14 | 3.65 | 70.16 | 76.66 | |
| HuatuoGPT-7BSize=7B2025.10 | 2.42 | 2.42 | 2.42 | 3.37 | 2.87 | 54 | 74.21 | |
| Qwen2.5VL-7BSize=7B2025.10 | 2.27 | 2.39 | 2.8 | 3.26 | 2.92 | 54.52 | 64.74 |