Benchmarks · knowledge
MMLU Pro (Vals)
Academic multiple-choice benchmark covering 14 subjects including STEM, humanities, and social sciences. Scores are the values published by VALS AI. Source benchmark version 1.
56 models · 56 results · updated 2026-09-01 · vals.ai ↗
Leaderboard
Fable 5.1 max
Heavy line: best score at each price. Thin lines join one model's effort levels. Cost uses each model's median subscription plan, or API price with no plan data. 37 more models under All models in settings.
More from this benchmark
15 published measuresTime per task137 results · lower is better
38 more models under All models in settings.
Score vs any published measure14 measures: sub-scores, splits, counts
Breakdown
darker is better within each column · click a column to sort| Model | ||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Fable 5.1 max | 96.1 | 95.2 | 93.8 | 94.9 | 95.1 | 90.8 | 86.7 | 85.3 | 87.5 | 97.3 | 89.9 | 90.2 | 96.7 | 93.7 |
| Opus 5 max | 95.7 | 93.5 | 93.3 | 93.9 | 93.7 | 89.9 | 87.8 | 86.6 | 86.6 | 97.1 | 88.0 | 89.0 | 94.8 | 92.5 |
| Fable 5 max | 93.6 | 93.2 | 93.9 | 94.1 | 93.8 | 90.0 | 84.6 | 85.3 | 86.5 | 97.6 | 89.8 | 90.6 | 95.8 | 92.2 |
| Gemini 3.8 Flash high | 95.0 | 93.8 | 92.5 | 92.7 | 92.1 | 88.8 | 84.5 | 83.7 | 82.0 | 96.2 | 87.6 | 88.2 | 94.8 | 91.5 |
| Gemini 3.7 Flash high | 94.7 | 92.8 | 92.3 | 92.0 | 91.7 | 89.3 | 84.7 | 84.0 | 81.9 | 96.0 | 87.6 | 89.4 | 94.3 | 91.1 |
| Opus 4.7 max | 94.6 | 93.3 | 93.3 | 92.7 | 92.3 | 88.0 | 82.4 | 83.2 | 82.1 | 95.9 | 86.9 | 89.0 | 93.4 | 91.2 |
| Opus 4.8 max | 94.7 | 92.6 | 92.8 | 92.7 | 91.8 | 87.7 | 85.1 | 82.7 | 81.9 | 95.9 | 86.1 | 86.6 | 93.5 | 90.0 |
| Gemini 3.5 Flash high | 93.9 | 93.3 | 93.1 | 91.7 | 91.9 | 87.8 | 83.6 | 81.6 | 80.4 | 96.3 | 87.6 | 87.6 | 94.1 | 90.4 |
| Grok 4.6 high | 95.0 | 92.6 | 92.0 | 91.7 | 91.8 | 85.8 | 84.4 | 81.1 | 81.8 | 95.9 | 87.2 | 89.2 | 92.1 | 90.9 |
| Qwen 3.7 Max | 95.1 | 93.5 | 92.3 | 91.2 | 91.7 | 85.8 | 84.4 | 82.2 | 79.5 | 96.0 | 86.0 | 89.2 | 93.5 | 90.0 |
| Gemini 3.6 Flash high | 94.7 | 92.8 | 92.3 | 91.7 | 90.5 | 86.7 | 84.6 | 83.5 | 80.1 | 96.0 | 86.5 | 87.6 | 93.5 | 89.5 |
| Grok 4.5 high | 94.3 | 93.3 | 91.6 | 91.0 | 91.4 | 87.2 | 84.0 | 81.4 | 82.0 | 95.9 | 87.2 | 87.6 | 92.0 | 90.2 |
| GPT-5.6 Sol max | 94.6 | 93.3 | 92.0 | 90.7 | 90.6 | 87.8 | 84.1 | 81.9 | 79.1 | 95.3 | 87.7 | 88.4 | 92.8 | 89.2 |
| Muse Spark 1.1 xhigh | 95.0 | 94.5 | 93.4 | 90.7 | 91.2 | 88.1 | 83.4 | 79.8 | 76.5 | 96.0 | 85.3 | 87.2 | 93.8 | 87.2 |
| Qwen 3.8 Max | 94.1 | 92.9 | 92.5 | 91.0 | 89.9 | 86.7 | 84.0 | 81.1 | 77.3 | 95.5 | 85.3 | 88.4 | 93.3 | 88.5 |
| Muse Spark 1.2 xhigh | 93.9 | 92.0 | 92.8 | 90.7 | 91.2 | 86.3 | 82.0 | 81.9 | 76.7 | 96.0 | 85.0 | 87.4 | 93.4 | 86.6 |
| GPT-5.5 xhigh | 94.0 | 92.0 | 91.3 | 89.8 | 89.9 | 84.8 | 84.4 | 81.6 | 76.9 | 94.7 | 86.8 | 87.8 | 91.5 | 88.5 |
| Kimi K3 max | 93.9 | 92.0 | 91.8 | 91.2 | 91.6 | 86.9 | 82.0 | 79.3 | 77.0 | 95.6 | 84.0 | 85.4 | 92.6 | 88.3 |
| Qwen 3.6 Plus max | 93.7 | 92.3 | 91.4 | 89.8 | 89.7 | 84.3 | 80.8 | 81.1 | 75.7 | 95.5 | 85.7 | 87.2 | 92.8 | 87.5 |
| Kimi K2.6 max | 93.7 | 91.6 | 92.5 | 91.2 | 89.9 | 87.1 | 80.7 | 78.5 | 74.3 | 95.6 | 85.0 | 84.6 | 92.1 | 89.2 |
| Sonnet 5 max | 93.6 | 91.6 | 90.8 | 89.0 | 89.5 | 81.6 | 82.3 | 80.3 | 78.7 | 95.9 | 85.4 | 86.6 | 91.5 | 88.7 |
| Sonnet 4.6 max | 92.7 | 91.9 | 91.9 | 90.0 | 90.6 | 85.2 | 82.4 | 81.1 | 73.6 | 94.1 | 83.1 | 86.2 | 92.6 | 87.3 |
| Muse Spark | 93.7 | 91.9 | 91.5 | 90.0 | 89.8 | 84.6 | 80.8 | 80.8 | 75.0 | 95.0 | 84.6 | 85.4 | 92.7 | 86.5 |
| MiniMax M2.1 | 93.6 | 90.1 | 91.3 | 87.1 | 90.3 | 81.8 | 83.7 | 79.5 | 78.9 | 92.7 | 86.0 | 86.0 | 89.5 | 88.0 |
| DeepSeek V4 Pro 0813 max | 93.7 | 91.0 | 91.7 | 90.0 | 89.2 | 83.4 | 81.3 | 78.5 | 74.9 | 95.1 | 84.5 | 84.6 | 92.5 | 87.2 |
All results
56 rows| # | Model | Effort | MMLU Pro (Vals) | Subscription $ / task | API $ / task | Time |
|---|---|---|---|---|---|---|
| 1 | Fable 5.1 | max | 92.4% | $0.015 | $0.098 | 21s |
| 2 | Opus 5 | max | 91.6% | $0.000 | $0.028 | 9.6s |
| 3 | Fable 5 | max | 91.5% | $0.009 | $0.082 | 25s |
| 4 | Gemini 3.8 Flash | high | 90.2% | $0.001 | $0.022 | 7.8s |
| 5 | Gemini 3.7 Flash | high | 90.1% | $0.001 | $0.011 | 4.2s |
| 6 | Opus 4.7 | max | 89.9% | $0.062 | $0.062 | 3.1 min |
| 7 | Opus 4.8 | max | 89.6% | $0.001 | $0.061 | 23s |
| 8 | Gemini 3.5 Flash | high | 89.5% | $0.025 | $0.025 | 26s |
| 9 | Grok 4.6 | high | 89.4% | $0.002 | $0.015 | 55s |
| 10 | Qwen 3.7 Max | max | 89.3% | $0.005 | $0.017 | 29s |
| 11 | Gemini 3.6 Flash | high | 89.3% | $0.000 | $0.013 | 7.3s |
| 12 | Grok 4.5 | high | 89.2% | $0.002 | $0.015 | 6.0 min |
| 13 | GPT-5.6 Sol | max | 89.1% | $0.001 | $0.032 | 16s |
| 14 | Muse Spark 1.1 | xhigh | 88.7% | $0.011 | $0.011 | 21s |
| 15 | Qwen 3.8 Max | max | 88.6% | $0.002 | $0.019 | 64s |
| 16 | Muse Spark 1.2 | xhigh | 88.3% | $0.005 | $0.011 | 44s |
| 17 | GPT-5.5 | xhigh | 88.1% | $0.001 | $0.028 | 42s |
| 18 | Kimi K3 | max | 88.0% | $0.002 | $0.021 | 39s |
| 19 | Qwen 3.6 Plus | max | 87.7% | $0.004 | $0.009 | 46s |
| 20 | Kimi K2.6 | max | 87.6% | $0.009 | $0.018 | 2.6 min |
| 21 | Sonnet 5 | max | 87.5% | $0.001 | $0.043 | 25s |
| 22 | Sonnet 4.6 | max | 87.3% | $0.066 | $0.066 | 55s |
| 23 | Muse Spark | 87.3% | $0.002 | $0.002 | 46s | |
| 24 | MiniMax M2.1 | 87.0% | $0.003 | $0.003 | 41s | |
| 25 | DeepSeek V4 Pro 0813 | max | 87.0% | $0.013 | $0.013 | 60s |
| 26 | GLM 5.1 | max | 86.9% | $0.003 | $0.012 | 69s |
| 27 | GLM 5.3 | max | 86.8% | $0.002 | $0.017 | 62s |
| 28 | GLM 5.2 | max | 86.7% | $0.000 | $0.014 | 52s |
| 29 | GPT-5.6 Terra | xhigh | 86.7% | $0.000 | $0.008 | 8.2s |
| 30 | Inkling | max | 86.3% | $0.004 | $0.040 | 73s |