SM Bench — Anti-hallucination
Safety · 2026-02-01
SM Bench Anti-hallucination tests whether models reject invented facts, acronyms, and unsupported premises instead of responding with fabricated certainty. Scores are difficulty-weighted judge credit across 100 fixed prompts.
Top models (higher is better)
| Model | Score |
|---|---|
| Grok 4.6 | 100.0 |
| GLM-5.3 | 100.0 |
| MiMo-V2-Pro | 100.0 |
| Opus 4.6 | 100.0 |
| Qwen3.7-Max (source unspecified) | 100.0 |
| Opus 4.8 | 100.0 |
| Sonnet 4.6 | 100.0 |
| GPT-6 Astra | 100.0 |
| Sonnet 5 | 100.0 |
| Claude Opus 5 | 99.0 |
| Claude Sonnet 5.5 | 99.0 |
| Gemini 3.1 Pro Preview | 99.0 |
| Claude Opus 5.5 | 99.0 |
| Opus 4.5 | 99.0 |
| Sonnet 4.5 | 99.0 |
| Kimi K3 | 99.0 |
| Muse Spark 1.2 | 99.0 |
| Grok 4.7 | 98.4 |
| MiMo-V2.6-Pro | 98.4 |
| Muse Spark 1.1 | 98.4 |
| MiMo-V2.5-Pro | 98.4 |
| Claude Fable 5 | 98.4 |
| Qwen3.5 397B A17B | 98.4 |
| Grok 4.5 | 97.9 |
| Grok 4.3 | 97.9 |
| Opus 4.7 | 97.4 |
| Gemini 3.7 Flash | 97.4 |
| Gemini 3 Pro Preview | 97.4 |
| MiMo-V2.6-Flash | 97.4 |
| GPT-6 Astra Pro | 97.4 |
| Grok 4.1 Fast | 96.9 |
| Gemma 4 31B IT | 96.9 |
| Qwen3.6 Plus Preview | 96.9 |
| GPT-6.1 Sol | 96.9 |
| Claude Fable 5.1 | 96.3 |
| GLM-5.2 | 96.3 |
| Gemini 3.8 Flash | 96.3 |
| Kimi K2.6 | 96.3 |
| MiniMax M3 | 96.3 |
| Claude Haiku 5.5 | 95.8 |