MMLU leaderboard
- GPT-4o leads
- 88.1%
- models measured
- 23
- lead over #2
- +1.7 pts
- median
- 80.0%
Leader on MMLU · read October 22
On October 22, 2026, 31 of our catalog's models had a MMLU result, and OpenAI GPT-4o led with 88.1%, ahead of Anthropic Claude 3.5 Sonnet at 87.3%.
Multiple-choice questions across 57 academic subjects; most current models are near the ceiling. Share answered correctly.
Role in the index
MMLU is shown as published and does not feed any of our indexes. Read the methodology
| # | Model | Result | Evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|
| 1 | GPT-4oOpenAI | 88.1% | Not benchmarked | $10.00 | |
| 2 | Claude 3.5 SonnetRetiredAnthropic | 87.3% | Not benchmarked | $15.00 | |
| 3 | Gemini 1.5 ProRetiredGoogle | 86.9% | Not benchmarked | $5.00 | |
| 4 | GPT-4OpenAI | 86.4% | Not benchmarked | $60.00 | |
| 5 | Llama 3.3 70B InstructMeta Llama | 86.3% | Not benchmarked | $0.72 | |
| 6 | Qwen2.5 72B InstructQwen | 85.3% | Not benchmarked | $0.40 | |
| 7 | Phi 4Microsoft | 84.8% | Not benchmarked | $0.14 | |
| 8 | Claude 3 OpusRetiredAnthropic | 84.6% | Not benchmarked | $75.00 | |
| 9 | GPT-4o (2024-08-06)OpenAI | 84.3% | Not benchmarked | $10.00 | |
| 10 | GPT-4o (2024-05-13)OpenAI | 84.2% | Not benchmarked | $15.00 | |
| 11 | Nova Pro 1.0Amazon | 82.0% | Not benchmarked | $3.20 | |
| 12 | GPT-4o-miniOpenAI | 81.8% | Not benchmarked | $0.60 | |
| 13 | GPT-4 TurboOpenAI | 81.3% | Not benchmarked | $30.00 | |
| 14 | Llama 3.1 70B InstructMeta Llama | 80.1% | Not benchmarked | $0.56 | |
| 15 | Mistral Large 2407Mistral AI | 80.0% | Not benchmarked | $6.00 | |
| 16 | Qwen2.5 Coder 32B InstructQwen | 79.1% | Not benchmarked | $1.00 | |
| 17 | Claude 2RetiredAnthropic | 78.5% | Not benchmarked | $24.00 | |
| 18 | Gemini 1.5 FlashRetiredGoogle | 77.9% | Not benchmarked | $0.30 | |
| 19 | Nova Lite 1.0Amazon | 77.0% | Not benchmarked | $0.24 | |
| 20 | GPT-4o-mini (2024-07-18)OpenAI | 76.7% | Not benchmarked | $0.60 | |
| 21 | Claude 3 SonnetRetiredAnthropic | 75.9% | Not benchmarked | $15.00 | |
| 22 | Gemma 2 27BGoogle | 75.7% | Not benchmarked | $0.65 | |
| 23 | Claude 3.5 HaikuRetiredAnthropic | 74.3% | Not benchmarked | $4.00 | |
| 24 | Claude 3 HaikuRetiredAnthropic | 73.8% | Not benchmarked | $1.25 | |
| 25 | Qwen2.5 7B InstructQwen | 72.9% | Not benchmarked | $0.20 | |
| 26 | Nova Micro 1.0Amazon | 70.8% | Not benchmarked | $0.14 | |
| 27 | GPT-3.5 Turbo (older v0613)OpenAI | 68.9% | Not benchmarked | $2.00 | |
| 28 | Mistral LargeMistral AI | 68.8% | Not benchmarked | $6.00 | |
| 29 | GPT-3.5 TurboOpenAI | 68.5% | Not benchmarked | $1.50 | |
| 30 | Mistral NemoMistral AI | 65.3% | Not benchmarked | $0.15 | |
| 31 | Llama 3.1 8B InstructMeta Llama | 56.1% | Not benchmarked | $0.08 |
Only models in our catalog are listed, and each appears once with its best published run. 215 rows were read from the source on October 22, 2026, and rows that match no catalog model are left out.