DSGym - MLE-Bench Lite - Above Median: leaderboard
MLE-Bench Lite's Kaggle competitions solved end to end, run by DSGym's default CodeAct agent (Nie et al., 2026) in an isolated Jupyter container at temperature 0 with all tools disabled. Share of competitions whose submission beats the leaderboard median, in percent; higher is better.
Metric: Above Median Rate (%). Source: arxiv.org. Saturation forecast: Around February 2027. 12 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.1 (High) | 45.45 | #225 (GPT-5.1) |
| 2 | Claude Sonnet 4.5 | 36.36 | #244 |
| 3 | GPT-5.1 (Medium) | 31.82 | #225 (GPT-5.1) |
| 4 | DeepSeek V3.1 | 27.27 | #500 |
| 5 | Kimi K2 0905 | 27.27 | #555 |
| 6 | GPT-5 (Medium) | 27.27 | #163 (GPT-5) |
| 7 | Claude Sonnet 4 | 22.73 | #397 |
| 8 | GPT-5.1 (Non-reasoning) | 22.73 | #225 (GPT-5.1) |
| 9 | Qwen 3 4B Instruct | 9.09 | #1660 |
| 10 | Qwen 2.5 7B Instruct | 0 | #2583 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=dsgym-mle-bench-lite-above-median · How It Works · Data refreshed daily, snapshot 2026-11-12.