MLE-bench leaderboard
As of November 8, 2026, Gemini 3.6 Flash (Google) leads MLE-bench at 63.9%, ahead of Gemini 3.5 Flash at 49.7% and Gemini 3.1 Pro Preview at 42.6%. 5 systems ranked. Every score here is published by the model's own lab; each row links to the page it comes from.
What is MLE-bench? MLE-bench tests whether an AI agent can do machine-learning engineering end to end on 75 real Kaggle competitions: prepare data, train models and submit. Made by OpenAI, October 2024.
Top models on MLE-bench, November 2026
| # | Model | Score | Price / 1M | Context | Source |
|---|---|---|---|---|---|
| 1 | Gemini 3.6 Flash Google | as reported | $0.75 / $3.75 | 1M | Vendor-reported Google DeepMind |
| 2 | Gemini 3.5 Flash Deprecated Google | as reported | $1.5 / $9 | 1M | Vendor-reported Google DeepMind |
| 3 | Gemini 3.1 Pro Preview Preview Google · listed as “Gemini 3.1 Pro” | as reported | $2 / $12 | 1M | Vendor-reported Google DeepMind |
| 4 | Gemini 3.5 Flash-Lite Google | as reported | $0.3 / $2.5 | 1M | Vendor-reported Google DeepMind |
| 5 | Gemini 3.1 Flash-Lite Deprecated Google | as reported | $0.25 / $1.5 | 1M | Vendor-reported Google DeepMind |
Can you trust MLE-bench scores?
No independent leaderboard publishes MLE-bench results for these models yet, so every number on this page is the lab's own report of its own model. Labs choose the harness, the effort level and the subset they report, and they rarely publish failures. Read the setting under each score before comparing two rows.
- The competitions are public Kaggle material, so models may have seen solutions in training.
- Compute and time budgets change the result; runs with different budgets are not comparable.
- The maintainers stopped taking leaderboard submissions in April 2026 while they work on comparability.
How MLE-bench works
The task
For each competition the agent gets the description and data in a container and has up to 24 hours to train, experiment and produce a submission file.
The score
The headline is the share of attempts that would earn a Kaggle medal, judged against each competition's private leaderboard.
Test set
75 Kaggle competitions, as published by the authors.
Made by
OpenAI, first published October 2024.
Official sources
- github.com/openai/mle-bench
- Paper: arxiv.org/abs/2410.07095
- Definition read on 2026-11-08
Sources and freshness
5 published scores from 2 sources. Each was read on its source page; the oldest reading on this page is from November 8, 2026. Prices beside each score come from our model price index and move when a vendor reprices.
- Vendor-reported Google DeepMind
- Vendor-reported Google DeepMind
MLE-bench FAQ
Answers derived from the 5 published scores on this page and the current price index.