MMLU Pro - Engineering
General QA · 2024-06-03
The Engineering 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 5.1 | 90.8 |
| Gemini 3.7 Flash | 89.3 |
| Gemini 3.8 Flash | 88.8 |
| Muse Spark 1.1 | 88.1 |
| Opus 4.7 | 88.0 |
| Gemini 3.1 Pro Preview | 88.0 |
| Gemini 3.5 Flash | 87.8 |
| GPT-5.6 Sol | 87.8 |
| Opus 4.8 | 87.7 |
| Gemini 3 Pro Preview | 87.6 |
| Gemini 3 Flash Preview | 87.3 |
| Grok 4.5 | 87.2 |
| Opus 4.6 | 87.1 |
| Kimi K2.6 | 87.1 |
| Kimi K3 | 86.9 |
| DeepSeek-V4-Pro | 86.7 |
| Gemini 3.6 Flash | 86.7 |
| Qwen3.8-Max | 86.7 |
| Opus 4.5 | 86.5 |
| Muse Spark 1.2 | 86.3 |
| Grok 4.6 | 85.8 |
| Inkling-Small | 85.8 |
| Qwen3.7-Max (source unspecified) | 85.8 |
| GPT-5.6 Luna | 85.7 |
| Inkling | 85.7 |
| Sonnet 4.6 | 85.2 |
| GPT-5.4 | 85.2 |
| Qwen3.5 Plus (2026-02-15) | 85.2 |
| GPT-5.5 | 84.8 |
| Muse Spark | 84.6 |
| Qwen3.6 Plus (2026-04-02) | 84.3 |
| Grok 4.3 | 84.2 |
| GLM-5.1 | 84.0 |
| GLM-5.3 | 84.0 |
| Kimi K2.5 | 83.8 |
| Gemini 3.1 Flash-Lite Preview | 83.7 |
| DeepSeek V4 Pro 0813 | 83.4 |
| GLM-5.2 | 83.4 |
| DeepSeek-V3.2 | 83.3 |
| DeepSeek-V4-Flash-0731 | 83.2 |