MiniCPM-2B
2B-parameter on-device LLM that matched Mistral-7B on public benchmarks.
About
MiniCPM-2B is a 2-billion-parameter language model from OpenBMB (Tsinghua NLP Lab and ModelBest), released on February 1, 2024. It was positioned as an end-side LLM: performance similar to Mistral-7B on public benchmarks, with better Chinese, math, and code ability, while outperforming much larger models like Llama2-13B, MPT-30B, and Falcon-40B.
The Apache 2.0 weights ship in bf16 and quantized GGUF forms, making it a popular choice for phones, laptops, and edge devices where a capable bilingual assistant must run fully offline.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 112; Trending score 80/100 from researcher assessment.