Baidu's Famo Agent 2.0 Tops OpenAI's MLE-Bench, Set for Official Release at Create 2026
Baidu Cloud has announced that its enterprise-level autonomous ML optimization agent — Famo Agent 2.0 (百度伐谋Agent 2.0) — has once again claimed the top spot on the MLE-Bench leaderboard, setting a new SOTA score. The official version will be released at Baidu's Create 2026 AI Developer Conference in May.
What is MLE-Bench?
MLE-Bench is a benchmark established by OpenAI containing 75 real-world engineering challenges sourced from Kaggle competitions. It tests AI agents' ability to perform end-to-end machine learning engineering tasks, including:
- Feature engineering and data preprocessing
- Model selection and hyperparameter tuning
- Ensemble construction
- Submission optimization
Key Details
- Developer: Baidu Cloud (百度智能云)
- Benchmark: MLE-Bench (OpenAI-established)
- Tasks: 75 real Kaggle competition problems
- Release: Official version at Create 2026 (May 2026)
- Category: Enterprise AI agent for automated ML
Significance
This achievement is notable for several reasons:
- Chinese AI competitiveness: Baidu's agent topping an OpenAI-designed benchmark demonstrates China's continued competitiveness in AI agent technology
- AutoML advancement: Famo Agent represents the cutting edge of automated machine learning, where AI agents autonomously solve complex engineering problems
- Enterprise readiness: Positioned as an enterprise-grade tool, not just a research demo
- Agentic AI trend: Part of the broader shift from standalone models to autonomous agent systems that combine LLMs with specialized tools
The MLE-Bench benchmark has become a key battleground for comparing AI agent capabilities across companies, with OpenAI, Google DeepMind, and now Baidu competing for the top position.