Phare Misinformation (English)
Safety · 2025-02-19
Phare task testing resistance to misinformation generation or endorsement. This row is the English language split.
Top models (higher is better)
| Model | Score |
|---|---|
| Haiku 4.5 | 99.1 |
| Opus 4.1 | 95.3 |
| Claude 3.5 Sonnet (Oct 2024) | 93.2 |
| Sonnet 4.5 | 91.9 |
| Claude 3.7 Sonnet | 91.6 |
| Opus 4.5 | 91.0 |
| Sonnet 5 | 91.0 |
| Llama 3.1 405B Instruct | 91.0 |
| Opus 4.6 | 88.5 |
| Haiku 3.5 | 87.0 |
| Kimi K2.6 | 85.7 |
| Command A | 84.8 |
| Qwen-Plus (2025-01-25) | 84.5 |
| Gemini 3.1 Pro Preview | 83.8 |
| Gemma 4 31B IT | 83.2 |
| Magistral Medium 1.2 | 83.2 |
| Grok 4 | 82.0 |
| Llama 3.3 70B Instruct | 81.7 |
| GPT-5 Mini | 79.2 |
| Sonnet 4.6 | 78.9 |
| GPT-4.1 Nano | 78.9 |
| Gemini 1.5 Pro | 78.6 |
| Gemini 2.5 Pro | 77.6 |
| GPT-5.1 | 77.0 |
| GPT-5 Nano | 77.0 |
| Gemini 2.0 Flash-Lite | 76.7 |
| Qwen3 Max (rolling alias) | 76.7 |
| Gemini 2.5 Flash-Lite | 76.4 |
| GPT-4o | 75.2 |
| Mistral Large 2 (Instruct 2407) | 75.2 |
| Gemini 2.0 Flash | 74.5 |
| Gemini 2.5 Flash | 74.2 |
| Llama 4 Maverick | 73.9 |
| Gemini 3.1 Flash-Lite | 73.3 |
| Kimi K2.5 | 72.4 |
| Grok 3 Mini | 71.4 |
| Llama 3.1 8B Instruct | 70.5 |
| Gemini 3.5 Flash | 69.6 |
| GPT-4.1 | 68.9 |
| DeepSeek-V3-0324 | 68.6 |