The AI community
building the future.
The leading open-source AI hub in China. Discover, share and deploy models, datasets and spaces. Accelerated downloads, localized docs, community-first.
Three hubs.
One ecosystem.
Models, Datasets and Spaces — everything the ML community needs to build, share and deploy AI, with China-optimized infrastructure.
Models
Discover, download and deploy 400,000+ pre-trained models — NLP, vision, audio, multimodal — from the global open-source community.
Datasets
Browse 100,000+ curated datasets for training and evaluation. Version-controlled, with built-in preview and streaming.
Spaces
Build and share interactive ML demos with Gradio, Streamlit, or Docker. Zero-config GPU hosting included.
Trending models
this week.
See what the community is building with. From large language models to image generators, the hottest open-source models — all accessible with accelerated downloads in China.
Qwen/Qwen2.5-72B-Instruct
THUDM/GLM-4-9B-Chat
BAAI/bge-large-zh-v1.5
stabilityai/stable-diffusion-xl
deepseek-ai/DeepSeek-V3
openai/whisper-large-v3
Fast access.
Local experience.
Accelerated model downloads via China CDN. Localized documentation, community forums, and enterprise support — built for the Chinese AI ecosystem.
CDN Accelerated
Download models at full speed through China-optimized CDN nodes.
Bilingual Docs
Complete Chinese documentation, tutorials, and API references.
Inference API
Run any model with a single API call — free tier included.
Enterprise Ready
Private model hosting, SSO, audit logs, and dedicated support for teams.
Clean energy powering the AI future.
Every inference run, every model download — powered by solar. Monarch Trace LLC is building AI infrastructure for a carbon-neutral civilization, starting in China.
Solar ComputeTrusted by teams across China
"We switched our entire pipeline to HuggingFace China. Model downloads went from hours to minutes with the China CDN. The localized API docs saved our team weeks of onboarding time."
"Spaces let us ship interactive demos to stakeholders same-day. The GPU-backed hosting means our diffusion models run smoothly without provisioning any infrastructure."
"The private Hub for our organization was set up in an afternoon. 200+ internal models with access control, versioning, and audit logging — exactly what enterprise AI teams need."
Start building.
Join the hub.
Create a free account to upload models, bookmark favorites, and join the discussion.