Machine Learning Engineering on MLE-Bench Lite
86.36Any Medal (%)Arbor
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| ArborModel=GPT-5.52026.06 | 86.36 | — | 77.27 | 100 | 95.45 | 0 | 9.09 | |
| AI-ScientistModel=Gemini-3-Flash2026.06 | 81.82 | — | 31.82 | 100 | 86.36 | 18.18 | 31.82 | |
| ArborModel=Gemini-3-Flash2026.06 | 81.82 | — | 40.9 | 100 | 86.36 | 13.64 | 27.27 | |
| AIBuildAIModel=Claude-Opus-4.62026.06 | 77.27 | — | 37.88 | 100 | 81.82 | 13.64 | 25.76 | |
| LoongFlowModel=Gemini-3-Flash2026.06 | 77.27 | — | 39.39 | 77.27 | 77.27 | 12.12 | 25.76 | |
| EvoMasterBackend Model=GPT-5.4, Maximum Runtime=24 hours2026.04 | 75.8 | 316 | — | — | — | — | — | |
| ML-Master 2.0Model=DeepSeekV3.2-Spe2026.06 | 75.76 | — | 30.3 | 100 | 84.85 | 13.64 | 31.82 | |
| Famou-Agent 2.0Model=Gemini-2.5-Pro2026.06 | 75.76 | — | 40.91 | 100 | 86.36 | 15.15 | 19.7 | |
| MARSModel=Gemini-3-Pro2026.06 | 74.24 | — | 53.03 | 100 | 89.39 | 6.06 | 15.15 | |
| DataMasterBackbone=GLM-52026.05 | 68.18 | — | 45.45 | — | — | — | — | |
| R&D-AgentModel=GPT-52026.06 | 68.18 | — | 33.33 | 77.27 | 74.24 | 12.12 | 22.73 | |
| LeerooModel=Gemini-3-Pro2026.06 | 68.18 | — | 30.3 | 68.18 | 68.18 | 18.18 | 19.7 | |
| CodexModel=GPT-5.5 (xhigh)2026.06 | 68.18 | — | 46.97 | 100 | 81.82 | 1.52 | 19.7 | |
| InternAgentModel=DeepSeek-R12026.06 | 62.12 | — | 34.85 | 100 | 78.79 | 10.61 | 16.67 | |
| DatasetResearcherBackbone=GLM-52026.05 | 59.09 | — | 27.27 | — | — | — | — | |
| AIRA-dojoModel=o32026.06 | 55 | — | 34.32 | 100 | 70.45 | 7.95 | 12.73 | |
| KompeteAILLM Engine=gemini-2.5-flash, Number of seeds=3, Time limit per run=6 hours2025.08 | 51.5 | — | — | — | — | — | — | |
| ML-MasterLLM Engine=deepseek-r12025.08 | 48.5 | — | — | — | — | — | — | |
| ML-MasterModel=DeepSeek-R12026.06 | 48.48 | — | 30.3 | 100 | 74.24 | 4.55 | 13.64 | |
| RD-AgentLLM Engine=o1-preview2025.08 | 48.18 | — | — | — | — | — | — | |
| AIDEModel=Gemini-3-Flash2026.06 | 45.45 | — | 31.82 | 77.27 | 54.55 | 4.55 | 9.09 | |
| MLE-STARLLM Engine=gemini-2.5-flash2025.08 | 43.9 | — | — | — | — | — | — | |
| ML-Master 2.0Backbone=GLM-52026.05 | 40.91 | — | 27.27 | — | — | — | — | |
| Claude CodeBackbone=GLM-52026.05 | 36.36 | — | 22.12 | — | — | — | — | |
| Initial ScoreDescription=ML-Master's code with data module removed2026.05 | 35.91 | — | 22.73 | — | — | — | — | |
| AIDELLM Engine=o1-preview2025.08 | 34.3 | — | — | — | — | — | — | |
| Codex*Backbone=GPT-5.2-Codex2026.05 | 22.73 | — | 18.18 | — | — | — | — | |
| OpenClawBackend Model=GPT-5.4, Maximum Runtime=24 hours2026.04 | 18.2 | — | — | — | — | — | — |