Open LLM Leaderboard v1 - TruthfulQA MC2: leaderboard
Original Hugging Face Open LLM Leaderboard zero-shot TruthfulQA multiple-choice MC2: the normalized probability mass assigned to all truthful answers, not MC1 accuracy or free-form answer grading. Percentage, higher is better. Frozen v1 archive; public questions may overlap model training.
Metric: MC2 (%) (0-shot). Source: huggingface.co. Saturation forecast: Estimated already saturated. 7162 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Myrrh_solar_10.7b_3.0 | 79.87 | #1060 |
| 2 | luxia-21.4B-alignment-v1.2 | 79.16 | #963 |
| 3 | DARE_TIES_13B | 78.66 | #856 |
| 4 | MixTAO-7Bx2-MoE-v8.1 | 78.57 | #815 |
| 5 | CarbonBeagle-11B-truthy | 78.55 | #1023 |
| 6 | YamshadowExperiment28-7B | 78.53 | #912 |
| 7 | MoE_13B_DPO | 78.47 | #857 |
| 8 | NeuralSynthesis-7B-v0.1 | 78.15 | #976 |
| 9 | NeuralSynthesis-7B-v0.3 | 78.13 | #903 |
| 10 | MultiverseEx26-7B-slerp | 78.12 | #911 |
| 11 | CognitiveFusion2-4x7B-BF16 | 78.12 | #926 |
| 12 | NeuralSynthesis-7B-v0.4-slerp | 78.07 | #996 |
| 13 | Truthful_DPO_TomGrc_FusionNet_7Bx2_MoE_13B | 78.02 | #848 |
| 14 | M7-7B | 77.93 | #887 |
| 15 | AlphaMonarch-7B | 77.91 | #1044 |
No result here: #3 Claude Opus 5.5, #5 GPT-6 Astra, #6 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=open-llm-leaderboard-v1-truthfulqa-mc2 · How It Works · Data refreshed daily, snapshot 2026-11-07.