Hallucination Suppression on HealthBench Hallu
2.37Refuted RateGPT-5.2-High
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GPT-5.2-High2026.02 | 2.37 | 2.78 | |
| Baichuan-M3-235BFact-Aware RL=true, Parameters=235B2026.02 | 2.45 | 2.07 | |
| Baichuan-M3Fact-Aware RL=false2026.02 | 4.68 | 3.64 | |
| Baichuan-M2-32BParameters=32B2026.02 | 5.73 | 5.43 |