MLE-Bench Lite
We show this table for reference; we do not rank on it.
A lightweight machine-learning competition benchmark that measures whether models can iteratively train, evaluate, and improve ML systems in low-resource settings.
Benchmark score on MLE-Bench Lite — November 13, 2026
We compile the MLE-Bench Lite rows from provider self-reports. Atria Dawn Preview leads the table at 86.2%, followed by MiniMax M2.7 (66.6%) and Sage-A1.5 (65.2%). We do not use these results to rank models overall.
Atria Dawn Preview
Shanghai Artificial Intelligence Laboratory
Reported by the labMiniMax M2.7
MiniMax
Reported by MiniMaxSage-A1.5
InternScience
Reported by InternScience4 modelsAgenticCurrentDisplay onlyUpdated November 13, 2026
Benchmark score table (4 models)
All 4 results on this page were reported by the labs that built the models.
| Rank | Model / configuration | Score | Who ran it | Parameters (B) | Open / closed |
|---|---|---|---|---|---|
| 1 | Atria Dawn PreviewShanghai Artificial Intelligence LaboratoryReported by the lab | 86.2% | Reported by the lab | Not reported | Open |
| 2 | MiniMax M2.7MiniMaxReported by MiniMax | 66.6% | Reported by MiniMax | Not reported | Open |
| 3 | Sage-A1.5InternScienceReported by InternScience | 65.2% | Reported by InternScience | Not reported | Open |
| 4 | Agents-A1-4BInternScienceReported by InternScience | 22.7% | Reported by InternScience | Not reported | Open |
- Lab-reported
- Benchmark-run
- Independent
Among the reported MLE-Bench Lite rows, Atria Dawn Preview is first at 86.2%. The third row is 21.0 points behind. The broader top-10 range is 63.5 points, so the table still separates the published systems.
4 models have been evaluated on MLE-Bench Lite. The benchmark falls in the Agentic category. MLE-Bench Lite is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.
About MLE-Bench Lite
Year
2026
Tasks
Low-resource ML competitions
Format
Autonomous iterative ML optimization
Difficulty
Agentic machine learning
MiniMax reports MLE-Bench Lite results from autonomous multi-round optimization on low-resource machine-learning competitions, making it a useful signal for agentic ML workflows.
Freshness and provenance
Version
MLE-Bench Lite 2026
Refresh cadence
Quarterly
Staleness state
Current
Question availability
Public benchmark set
BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.
Questions
What does MLE-Bench Lite measure?
A lightweight machine-learning competition benchmark that measures whether models can iteratively train, evaluate, and improve ML systems in low-resource settings.
Which model scores highest on MLE-Bench Lite?
Atria Dawn Preview by Shanghai Artificial Intelligence Laboratory currently leads with a score of 86.2% on MLE-Bench Lite.
How many models are evaluated on MLE-Bench Lite?
4 AI models have published results on MLE-Bench Lite in the BenchLM catalog.
Compare top models on MLE-Bench Lite
Know when it’s worth switching models
The model to choose, the cheaper alternative, and the release we would wait on.
Read a sample issueJoin 7,000+ subscribers.
One email each week. Unsubscribe anytime.