From the model card
Fields below are copied from the tags and counters on the HuggingFace repository openbmb/MiniCPM5-1B at our last fetch. They are set by the uploader, not verified by us; rows with no tag are omitted. How this page is made.
- Publisher (HF namespace)
- openbmb
- Pipeline tag
- text-generation
- Library
- Transformers
- Weight formats
- safetensors
- License tag
apache-2.0— read the license file in the repo before relying on it- Language tags
- English (en), Chinese (zh)
- Papers cited
- arXiv:2506.07900, arXiv:2602.09003, arXiv:2512.16649, arXiv:2604.13016
- Datasets declared
- openbmb/Ultra-FineWeb, openbmb/Ultra-FineWeb-L3, openbmb/UltraData-Math, openbmb/UltraData-SFT-2605
- Downloads (HF counter at last fetch)
- 744,505
- Likes (HF counter at last fetch)
- 1,099
- Model card
- https://huggingface.co/openbmb/MiniCPM5-1B
Use cases
- On-device assistant applications on mobile and edge hardware
- Low-power IoT or embedded inference
- Long-context document summarization within CPU constraints
- Tool-calling agents on edge devices
- Chinese-English bilingual edge assistants
Pros
- Apache 2.0 license
- 1B scale enables CPU or NPU inference on mobile hardware
- Tool-calling capability at edge scale is rare among open models
- Long-context support beyond most 1B models
- 829 likes reflects strong community validation
Cons
- 1B parameters lag 7B+ models significantly on complex reasoning
- Long-context performance degrades more sharply than larger models
- on-device optimization may need target-platform specific compilation
- Chinese-English bilingual focus limits multilingual coverage
Tags
transformerssafetensorsllamatext-generationminicpmminicpm5long-contexttool-callingon-deviceedge-aiconversationalenzhdataset:openbmb/Ultra-FineWebdataset:openbmb/Ultra-FineWeb-L3dataset:openbmb/UltraData-Mathdataset:openbmb/UltraData-SFT-2605arxiv:2506.07900arxiv:2602.09003arxiv:2512.16649