MLS-Bench (Agent) - Deep Learning: leaderboard
MLS-Bench (Agent) Deep Learning scores the final submission after up to 20 actions and three test runs on the 11 Deep Learning tasks of MLS-Bench (2026), where a model must invent a better ML method inside a frozen research codebase. Task scores are normalized so the worst reproduced human baseline is 0 and the best is 50, on a 0 to 100 scale where higher is better.
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 11 MLS-Bench tasks of the Deep Learning 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 March 2028. 5 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | High | 44.8 | #99 |
| 2 | Claude Opus 4.6 | Thinking | 38.5 | #113 |
| 3 | DeepSeek V3.2 | Thinking | 33.7 | #419 |
| 4 | GPT-5.4 | High | 31.6 | #135 |
| 5 | Qwen 3.6 Plus | Thinking | 30.9 | #238 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=mls-bench-agent-deep-learning · How It Works · Data refreshed daily, snapshot 2026-11-21.