MiniCPM5<\/strong> is OpenBMB’s series of small language models, designed from the start to run on-device<\/strong>\u2014on phones and laptops. Following the 1B model, the second release is MiniCPM5-2B<\/strong>, a dense model with about 2.5B parameters.<\/p>\n The architecture is a plain Llama-style model ( The publisher bills it as the best open model in the 2B class, with an average score of 53.9<\/strong> across its comparison table. That is far ahead of the 2B-class LFM2.5-2.6B (33.2) and Qwen3.5-2B (28.0), and higher than Qwen3.5-4B (51.1), a model twice its size.<\/strong> Broken down by benchmark, its strengths and weaknesses are clear:<\/p>\n \u2192 Scroll horizontally to see all columns<\/p>\nLlamaForCausalLM<\/code>) with a 131,072-token context window. Because it needs no special implementation, common tools such as llama.cpp, Ollama and LM Studio handle it as-is. The publisher itself distributes GGUF, MLX (for Apple Silicon) and GPTQ (4-bit)<\/strong> builds, plus a draft model for faster inference (DSpark) and intermediate training checkpoints (Base, Midtrain, SFT).<\/p>\nWhat Makes It Stand Out<\/h2>\n