Phare Tools Reliability (French)
Safety · 2025-03-27
Phare task testing reliability when using or reasoning about tool/plugin outputs. This row is the French language split.
Top models (higher is better)
| Model | Score |
|---|---|
| Sonnet 4.6 | 98.8 |
| Opus 4.6 | 97.7 |
| Gemini 3.1 Pro Preview | 96.4 |
| Qwen3.7 Max Preview | 96.2 |
| Sonnet 5 | 95.3 |
| DeepSeek-V4-Flash | 95.3 |
| DeepSeek-V4-Pro | 95.2 |
| Gemini 3.5 Flash | 94.9 |
| Sonnet 4.5 | 94.1 |
| GLM-5.2 | 94.1 |
| Opus 4.5 | 93.3 |
| GPT-5.5 | 93.2 |
| Claude 3.5 Sonnet (Oct 2024) | 92.7 |
| Qwen3.7-Plus | 92.7 |
| Grok 4 | 92.6 |
| Opus 4.1 | 92.0 |
| Gemini 3 Pro Preview | 90.9 |
| Kimi K2.5 | 90.8 |
| Gemma 4 31B IT | 90.1 |
| Mistral Medium 3.1 | 90.0 |
| Mistral Medium 3.5 | 89.6 |
| Haiku 4.5 | 88.9 |
| Grok 3 | 88.9 |
| Grok 4.3 | 88.8 |
| Mistral Large 2 (Instruct 2407) | 88.6 |
| GPT-5.1 | 88.1 |
| Haiku 3.5 | 88.0 |
| Mistral Large 3 675B Instruct 2512 | 87.5 |
| Gemini 3.1 Flash-Lite | 87.3 |
| GPT-4.1 Mini | 86.7 |
| Mistral Small 3.1 24B Instruct 2503 | 86.6 |
| GPT-4.1 | 86.3 |
| GPT-5 Mini | 85.3 |
| GPT-4o | 85.1 |
| GPT-5.2 | 84.9 |
| GPT-5 Nano | 84.9 |
| gpt-oss-120b | 84.0 |
| Kimi K2.6 | 84.0 |
| Claude 3.7 Sonnet | 83.7 |
| Grok 4 Fast | 83.0 |