ModelScope
An open-source AI model community platform initiated by Alibaba, providing model hosting and download services for the Chinese market. It aggregates mainstream open-source models including DeepSeek and Zhipu GLM, with servers located in China to provide favorable download speeds for Chinese users.
Core Facts
Qwen-Image-2.1 is available through multiple channels including GitHub, ModelScope, and Hugging Face.
DiffSynth-Studio is an open-source diffusion model tooling project developed by the ModelScope team.
ModelScope（魔搭社区）是阿里巴巴达摩院推出的国内AI模型托管平台，提供模型托管、下载加速和在线体验功能
Timeline (last 90 days)
The author of Qwenputer open-sourced the model weights on Hugging Face under the name Qwen3.5-4B-Qwenputer, and provided an online demo on ModelScope.
OXR's repository is at GitHub's SoMarkAI/OXR, the weights are at Hugging Face's SoMarkAI/OXR-1.0, and it is released simultaneously on ModelScope; the license is Apache 2.0 with additional terms, including threshold restrictions on commercial use.
DeepSeek 4.1 Flash's open-source weights can be downloaded on Hugging Face and ModelScope, the community has already released quantized versions, and the most compressed version is greatly reduced in size but still requires high-end GPUs.
ModelScope was initiated by Alibaba DAMO Academy and is a mainstream domestic model hosting and open-source community; HF Mirror is the domestic mirror site of Hugging Face; Moda Community is a model-sharing platform for domestic developers.
On the model wizard page, the enhanced edition offers a dropdown to switch to Moda Community, ModelScope, the HF Mirror domestic mirror, or the official Hugging Face download source.
The model has been open-sourced on Hugging Face and ModelScope, totaling about 510GB, and the community has already released quantized versions (e.g., Unsloth's 4bit GGUF at only 233GB).
ModelScope announced the open-sourcing of Spark X 2.64B, a 4-billion-parameter model supporting a 1-million-token context and offering three modes: no-thinking, low-budget, and high-budget.
Strata Launcher pulls model weights from the domestic ModelScope community to avoid unstable Hugging Face connections.
Both the FP8 and BF16 checkpoints of Step 5 Preview are now available on Hugging Face and ModelScope.
Among official channels, Hugging Face contributed 985,975 cumulative downloads with 12 model repositories, while the total cumulative downloads of the remaining 5 platforms (Ollama 15,851, ModelScope 11,116, Modelers 5,210, AtomGit 1,513, SCNet 16) are less than 34,000.
15 more timeline events
All Facts (20)
Qwen-Image-2.1 is available through multiple channels including GitHub, ModelScope, and Hugging Face.
65%VerifiedDiffSynth-Studio is an open-source diffusion model tooling project developed by the ModelScope team.
65%VerifiedModelScope（魔搭社区）是阿里巴巴达摩院推出的国内AI模型托管平台，提供模型托管、下载加速和在线体验功能
65%UnverifiedQwen3 27B uses the Apache 2.0 license and is available on Hugging Face and the ModelScope community for free download, deployment, and commercial use.
60%UnverifiedOXR's repository is at GitHub's SoMarkAI/OXR, the weights are at Hugging Face's SoMarkAI/OXR-1.0, and it is released simultaneously on ModelScope; the license is Apache 2.0 with additional terms, including threshold restrictions on commercial use.
50%UnverifiedDeepSeek 4.1 Flash's open-source weights can be downloaded on Hugging Face and ModelScope, the community has already released quantized versions, and the most compressed version is greatly reduced in size but still requires high-end GPUs.
50%UnverifiedOn the model wizard page, the enhanced edition offers a dropdown to switch to Moda Community, ModelScope, the HF Mirror domestic mirror, or the official Hugging Face download source.
50%UnverifiedModelScope was initiated by Alibaba DAMO Academy and is a mainstream domestic model hosting and open-source community; HF Mirror is the domestic mirror site of Hugging Face; Moda Community is a model-sharing platform for domestic developers.
50%UnverifiedThe model has been open-sourced on Hugging Face and ModelScope, totaling about 510GB, and the community has already released quantized versions (e.g., Unsloth's 4bit GGUF at only 233GB).
50%UnverifiedStrata Launcher pulls model weights from the domestic ModelScope community to avoid unstable Hugging Face connections.
50%UnverifiedVLM-Seek's inference code can be pulled from the official GitHub repository, while the model weights and region detector weights can be downloaded from the ModelScope community.
50%UnverifiedThe Qwen series is simultaneously released and maintained on both Hugging Face and ModelScope platforms.
50%UnverifiedAlibaba releases model weights through Hugging Face and ModelScope, using a license that is relatively friendly to commercial use.
50%UnverifiedModelScope is one of the main distribution channels for the Qwen series of models.
50%UnverifiedHF Mirror (hf-mirror.com) is a mirror site maintained by the domestic community to address the slow speed of directly connecting to the Hugging Face main site from within China.
50%UnverifiedMiniCPM5-2B is available for download via ModelScope.
50%UnverifiedModelScope has established its own MCP Plaza, which hosts services provided by a large number of third-party ecosystem companies.
50%UnverifiedAlmost all models in GGUF format on ModelScope and Hugging Face can run directly on Llama.cpp.
50%UnverifiedIn the command for downloading a single file using ModelScope, the model parameter corresponds to the model repository name, the filename corresponds to the specific model file, and the download directory is represented by ./ for the current folder.
50%UnverifiedThe author of Qwenputer open-sourced the model weights on Hugging Face under the name Qwen3.5-4B-Qwenputer, and provided an online demo on ModelScope.
50%