Knowledge & reasoning
MMLU-Pro leaderboard
As of 2026-10-07, the highest measured score on MMLU-Pro is 89.8% by Gemini 3 Pro Preview (high).
Measured scores: artificial-analysis.
| # | Model | Score | Source |
|---|---|---|---|
| 1 | Gemini 3 Pro Preview (high) | 89.8% | measured |
| 2 | Gemini 3 Pro Preview (low) | 89.5% | measured |
| 3 | Claude Opus 4.5 | 89.5% | measured |
| 4 | Gemini 3 Flash | 89.0% | measured |
| 5 | Claude Opus 4.5 (non-reasoning) | 88.9% | measured |
All leaderboards
Agentic · terminal
Agentic · tools
- AA-AnalystAgent
- APEX-Agents
- AutomationBench
- EnterpriseOps-Gym
- Harvey LAB
- ITBench SRE
- τ³-Bench Banking
- τ²-Bench Telecom