MiniCPM5 (1B / 2B)
🏆 My role: Agent Capability Lead
6.9M+ Downloads
MiniCPM5 is an open-source LLM series built for on-device, local, and resource-constrained deployment, with a native 128K context and out-of-the-box support in mainstream inference engines. I led the end-to-end construction of its agentic capabilities: tool-calling and multi-turn interaction data, agentic reinforcement learning, and the evaluation suite.
Key results:
- 🥇 The 2B model leads its size class on agentic benchmarks such as BFCL v4, τ²-Bench, SWE-bench Verified, and GAIA, and its overall average surpasses the larger Qwen3.5-4B
- 📦 MiniCPM5-1B and MiniCPM5-2B (including GGUF / MLX variants) have each reached about 3.45M cumulative downloads on Hugging Face — over 6.9M in total
- 🔌 Ships with an OpenAI-compatible tool-calling parser and Agent Skills for coding agents
Authors
Haotian Chen
(he/him)
Research Assistant Professor · LLM Post-Training / AI4AI
I am a Research Assistant Professor at the School of Artificial Intelligence, Shanghai Jiao Tong University, where I work with Prof. Junchi Yan at RethinkLab. I study how to build AI systems that can automate long-horizon, effort-intensive, and creativity-demanding tasks such as research, engineering, and development. My current work focuses on LLM post-training for autonomous agents, agent memory and continual learning, and AI4Research / AI4AI. Before joining SJTU, I received my PhD in Data Science from Fudan University, advised by Prof. Xiangdong Zhou, and completed postdoctoral research at Tsinghua University (THUNLP), working with Prof. Zhiyuan Liu and Prof. Maosong Sun. I was also a research intern at the Machine Learning Research Group of Microsoft Research Asia, mentored by Xiao Yang, and at the Institute for Interdisciplinary Information Sciences (IIIS), Tsinghua University, working with Prof. Yang Yu.