MLS-Bench (Agent) - Language Models: leaderboard
Metric: Mean normalized task score (0-100): each metric is anchored so the worst reproduced human baseline scores 0 and the best 50 (a metric's theoretical optimum maps to 100), metrics are averaged with human weights within a setting and by geometric mean across at least three generalization settings, over the 18 MLS-Bench tasks of the Language Models area; Agent protocol: the model's final submission after at most 20 actions including 3 test calls; high reasoning effort with a 10,000-token thinking budget, web search disabled, fixed seeds; higher is better. Source: arxiv.org. Saturation forecast: Around 2031. 5 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) (High) | 43.8 |
| 2 | DeepSeek V3.2 (Thinking) | 41.1 |
| 3 | Qwen 3.6 Plus (Thinking) | 38.1 |
| 4 | Claude Opus 4.6 (Thinking) | 35 |
| 5 | GPT-5.4 (High) | 34.7 |
Interactive version: theaggregate.ai/benchmark?slug=mls-bench-agent-language-models · How It Works · Data refreshed daily, snapshot 2026-10-07.