Healthcare QA on HealthBench
41.45ScoreARES-RL
Evaluation Results
| Method | Links | |
|---|---|---|
| ARES-RLBackbone=Qwen3-32B2026.05 | 41.45 | |
| WebscaleBackbone=Qwen3-32B2026.05 | 36.08 | |
| ARES-SFTBackbone=Qwen3-32B2026.05 | 35.78 | |
| CPTBackbone=Qwen3-32B2026.05 | 35.04 | |
| NaturalReasoningBackbone=Qwen3-32B2026.05 | 32.94 |