Phare Factuality (Spanish)
Safety · 2025-02-19
Phare task testing factual answer reliability and hallucination resistance. This row is the Spanish language split.
Top models (higher is better)
| Model | Score |
|---|---|
| Gemini 3.1 Pro Preview | 85.7 |
| Gemini 3 Pro Preview | 84.8 |
| Gemini 3.5 Flash | 81.9 |
| GPT-5.5 | 80.0 |
| GPT-5.1 | 78.1 |
| Gemini 2.5 Pro | 77.1 |
| GPT-5 | 75.2 |
| GPT-5.2 | 75.2 |
| Opus 4.6 | 73.3 |
| Grok 4 | 72.4 |
| Gemini 3.1 Flash-Lite | 71.4 |
| GLM-5.2 | 70.5 |
| GPT-4.1 | 69.5 |
| Sonnet 4.6 | 68.6 |
| Qwen3.7 Max Preview | 68.6 |
| Grok 3 | 67.6 |
| Kimi K2.6 | 66.7 |
| Opus 4.5 | 65.7 |
| Opus 4.1 | 64.8 |
| DeepSeek-V4-Pro | 64.8 |
| Claude 3.5 Sonnet (Oct 2024) | 63.8 |
| Grok 4.3 | 63.8 |
| Claude 3.7 Sonnet | 62.9 |
| DeepSeek-R1-0528 | 62.9 |
| Kimi K2.5 | 62.9 |
| Sonnet 4.5 | 61.0 |
| DeepSeek-V4-Flash | 61.0 |
| Gemini 2.0 Flash | 60.0 |
| GPT-4o | 60.0 |
| Mistral Large 3 675B Instruct 2512 | 59.0 |
| Gemini 2.5 Flash | 58.1 |
| DeepSeek-V3 | 57.1 |
| DeepSeek-V3-0324 | 57.1 |
| Sonnet 5 | 56.2 |
| Qwen3 Max (rolling alias) | 56.2 |
| Llama 4 Maverick | 55.2 |
| Gemini 1.5 Pro | 53.3 |
| Grok 3 Mini | 52.4 |
| Mistral Medium 3.1 | 52.4 |
| Mistral Large 2 (Instruct 2407) | 51.4 |