China LLMLandscape
Explore the Chinese model ecosystem across language, multimodal, embedding, coding, and AI agent frameworks, with a clear view of leading builders and releases.
As AI advances rapidly, large models have become a key driver of frontier innovation. Chinese companies and research organizations are moving quickly to build and release a wide range of models. Whether the model is open source or offered as a closed API service, tracking the domestic ecosystem and its changes in depth has become increasingly important.
Landscape
China LLM ecosystem relationship map
A panoramic view of major players in the China LLM ecosystem and how they connect. Drag and zoom are supported, and a larger desktop screen works best.
China LLM parameter size comparison
Models are sorted by parameter size. Hover to inspect more details.
Parameter size bands
Organizations
The builders behind China LLMs
Track leading Chinese technology companies and research institutions and their latest open ecosystem contributions.
阿里巴巴
Tracked 125 models
智谱AI
Tracked 43 models
DeepSeek-AI
Tracked 40 models
百度
Tracked 19 models
北京智源人工智能研究院
Tracked 16 models
MiniMaxAI
Tracked 11 models
腾讯AI实验室
Tracked 10 models
上海人工智能实验室
Tracked 10 models
百川智能
Tracked 10 models
零一万物
Tracked 10 models
Moonshot AI
Tracked 10 models
字节跳动Seed团队
Tracked 8 models
元象XVERSE
Tracked 7 models
面壁智能
Tracked 5 models
Tencent ARC
Tracked 5 models
华为
Tracked 4 models
IDEA研究院
Tracked 4 models
StepFunAI
Tracked 4 models
昆仑万维
Tracked 3 models
达摩院
Tracked 2 models
中科闻歌
Tracked 2 models
猎户星空科技
Tracked 2 models
浙江大学
Tracked 1 models
Benchmarks
Benchmark snapshots for China LLMs
Follow mainstream benchmark results such as MMLU, C-Eval, and GSM8K for a more objective multi-dimensional view of model capability.
Covers core skills such as language understanding, knowledge QA, and reasoning to show each model’s broad capability profile.
View details
Focused on code-generation suites such as HumanEval and MBPP to compare programming assistance and generation quality.
View details