OpenBMB/VisCPM

Health v2 · maintenance only

Flags: no_releases no_license

Adoption not part of the score

1062 stars · 88 forks observed · 2026-08-28

What it is AI-extracted, prompt v1, taxonomy v1, 2026-08-30, confidence not recorded

VisCPM is a family of open-source bilingual (Chinese/English) multimodal large models built on the 10B CPM-Bee language model, comprising VisCPM-Chat for image-to-text multimodal conversation and VisCPM-Paint for text-to-image generation. It fuses a Muffin visual encoder and a Diffusion-UNet decoder and achieved spotlight status at ICLR 2024.

Use cases

• build a multimodal chatbot that understands images in Chinese and English
• generate images from Chinese text prompts
• run bilingual vision-language question answering
• research cross-lingual transfer in multimodal models
• deploy a web demo for image conversation
• integrate multimodal model inference via API

When to choose

• you need strong Chinese-language multimodal understanding or generation
• you want open-source image-to-text and text-to-image models from one family
• you are researching bilingual generalization in multimodal LLMs

When to avoid

• you need the latest state-of-the-art multimodal performance (the team recommends MiniCPM-V or OmniLMM)
• you require a permissive license for commercial use (no license is specified)
• you lack GPU resources for 10B-parameter model inference

Facets

library · maturity maintenance

machine-learning deep-learning llm-inference image-processing chatbot stable-diffusion large-language-models computer-vision artificial-intelligence image-processing python multimodal text-to-image bilingual chinese vision-language-model cpm-bee iclr-2024 natural-language-processing gpu linux docker

Member repositories

For agents

markdown · JSON · MCP: product_card(name="OpenBMB/VisCPM")