MMLU Pro - Philosophy
General QA · 2024-06-03
The Philosophy split of MMLU Pro. This child benchmark separates a source-reported subtask or subtrack from the parent aggregate so scores at different grains do not share one benchmark_slug.
Top models (higher is better)
| Model | Score |
|---|---|
| Claude Fable 5 | 90.6 |
| Gemini 3.1 Pro Preview | 90.4 |
| Opus 4.5 | 89.4 |
| Qwen3.7-Max | 89.2 |
| Opus 4.7 | 89.0 |
| Claude Opus 5 | 89.0 |
| Gemini 3 Pro Preview | 89.0 |
| Opus 4.6 | 88.4 |
| GPT-5.6 Sol | 88.4 |
| Opus 4.1 | 88.2 |
| GPT-5.5 | 87.8 |
| Gemini 3.5 Flash | 87.6 |
| Gemini 3.6 Flash | 87.6 |
| Grok 4.5 | 87.6 |
| Muse Spark 1.1 | 87.2 |
| Qwen3.6 Plus (2026-04-02) | 87.2 |
| GPT-5.4 | 87.0 |
| Grok 4.3 | 86.8 |
| Opus 4.8 | 86.6 |
| Sonnet 5 | 86.6 |
| Sonnet 4.6 | 86.2 |
| GPT-5 | 86.2 |
| Sonnet 4.5 | 86.0 |
| GPT-5.1 | 86.0 |
| MiniMax M2.1 | 86.0 |
| Opus 4 | 85.6 |
| GPT-5.2 | 85.6 |
| Qwen3.5 Plus (2026-02-15) | 85.6 |
| Gemini 3 Flash Preview | 85.4 |
| Kimi K3 | 85.4 |
| Muse Spark | 85.4 |
| o3 | 85.2 |
| DeepSeek-V4-Pro | 84.6 |
| Kimi K2.6 | 84.6 |
| GLM-5.2 | 84.2 |
| GPT-5.6 Terra | 84.0 |
| GLM-5.1 | 83.9 |
| Grok 4 | 83.4 |
| Grok 4.20 | 83.4 |
| GLM-5 | 83.2 |