Gitee AI 模力方舟
OSChina's serverless hub for open model inference, tuning and apps
Overview
Gitee AI (模力方舟) is Open Source China's serverless platform for running, fine-tuning and deploying open models like Qwen and DeepSeek, with an app marketplace and a GiEngine inference engine tuned for domestic GPUs. Aimed at Chinese developers wanting a one-stop model shop.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 12 Jul 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
What you can produce with Gitee AI 模力方舟
- Serverless inference API compatible with the OpenAI SDK
- 100 free daily calls per model
- Model fine-tuning via a browser-based Jupyter environment
- One-click Streamlit/Gradio app hosting and Dockerfile-based deployment
- GiEngine inference engine optimized for domestic GPU/NPU hardware
- App marketplace for publishing and monetizing AI apps
- Integrations with Dify, n8n, ComfyUI and Claude Code
ASEAN Perspective
Gitee AI 模力方舟 in Southeast Asia
ASEAN-region availability and pricing notes coming soon. Drop the editorial team a note via /contact/ if you can supply local context (Singapore/Malaysia/Indonesia/Thailand/Vietnam).
Gitee AI's pitch is simple: take Open Source China's existing developer audience — Gitee is China's answer to GitHub — and layer a model marketplace on top, with a 100-calls-a-day free quota and pay-per-call pricing after that. The model catalog leans on the usual open-weight names (Qwen, DeepSeek, ChatGLM, Stable Diffusion) rather than anything proprietary, and integrations with Dify, n8n, ComfyUI and Claude Code make it easy to drop in as an OpenAI-compatible endpoint.
The catch is that it's one of at least three near-identical Chinese serverless-inference platforms (alongside PPIO and SiliconFlow) competing mostly on price and GPU availability, with little to differentiate Gitee AI besides its Gitee-community distribution. Fine for a Chinese developer already living in the Gitee ecosystem; less compelling for anyone outside it.
What people say
Gitee AI launched into open beta at the end of December 2023, built by Open Source China (OSChina), the company behind the Gitee code-hosting platform — effectively China's GitHub, with a large existing base of registered repositories and developers. 模力方舟 ("Moli Fangzhou", roughly "AI Ark") bolts a serverless model marketplace onto that existing developer base rather than starting from zero.
The core offering is straightforward: browse a model library spanning text, image, speech and vision models — Qwen2.5-72B, ChatGLM, Stable Diffusion, Flux and ChatTTS all show up in the catalog — and call them through an OpenAI-compatible API. New accounts get 100 free calls per day per model, enough to prototype before switching to metered pricing or a prepaid token/model resource package. Beyond inference, the platform includes a browser-based Jupyter environment, Streamlit/Gradio app hosting, and Dockerfile-based deployment, so a model can go from tested to a shipped demo app without leaving the site.
Its most-cited moment is being an early, low-friction host for DeepSeek's models when they went viral in China in early 2025 — developers reached for Gitee AI's free daily quota to try DeepSeek without paying for API access elsewhere. Chinese AI-tool directories and aggregator sites consistently describe it as cost-effective and note the appeal of Gitee's existing trust among Chinese developers, but independent user reviews — G2-style ratings, Reddit threads, Trustpilot — are essentially nonexistent outside China; this is a domestic-market product with domestic-market word of mouth.
Where it clearly overlaps with rivals: PPIO and SiliconFlow both offer near-identical OpenAI-compatible serverless catalogs at similar prices, and none of the three has an obvious technical moat beyond GPU sourcing and community trust. Gitee AI's edge, such as it is, comes from OSChina's existing developer relationships rather than differentiation on the models themselves — the GiEngine inference engine it promotes for domestic accelerators is a genuine point of difference for teams that can't or won't use Nvidia GPUs, but that's a narrow audience outside China.
Bottom line from what's publicly documented: reliable, cheap and easy to integrate for a China-based developer already inside the Gitee ecosystem; there's no strong reason for an ASEAN or global team to pick it over SiliconFlow or a Western inference host unless domestic compute compliance specifically matters.
Summary of public user & expert reviews, compiled by RECATOOLS.
About this listing
This entry was compiled from publicly available data including Gitee AI 模力方舟's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with Gitee AI 模力方舟 unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to Gitee AI 模力方舟 directly →
Spotted something out of date? Suggest an update →
Alternatives to Gitee AI 模力方舟
More in LLMs & Chat
