Chinese AI models

List of Chinese AI models

LLMs

LLM Name	Release	Developer	Description	Lang	Links
Qwen 2.5-Max	29.01.25	Alibaba Cloud	This model is the latest version of Qwen, featuring enhanced multimodal capabilities and improved performance for both image and text understanding tasks. It supports a wide range of applications including chatbots, content generation, and more.	RU, EN, CN	Homepage HuggingFace GitHub Paper Chat With News
DeepSeek-R1	20.01.25	Deepseek	We introduce our first-generation reasoning models, DeepSeek-R1-Zero and DeepSeek-R1.	RU, EN, CN	Homepage HuggingFace GitHub Paper Chat With News
Qwen1.5	06.02.24	Alibaba Cloud	Qwen1.5 is the improved version of Qwen, the large language model series.		Homepage HuggingFace GitHub
Tiangong 2.0 MoE	06.02.24	Kunlun Wanwei	It adopts the industry's top MoE expert hybrid model architecture and has stronger ability to handle complex tasks. Supports ultra-long context windows of up to 100K. Probably some specific access is required for non China users.	CN	Homepage News
deepseek-moe, -coder, -llm, etc	01.12.23	Deepseek	It has been trained from scratch on a vast dataset of 2 trillion tokens in both English and Chinese.	RU, CN, EN	HuggingFace GitHub arXiv Chat With
iFlytekSpark-13B-base, -chat, -Lora, -Character	30.01.24	iFLYTEK	According to the tests by iFlytek, it is 20%+ ahead of other open source models of the same size in many aspects.		Gitee News
MiniCPM-2B	01.02.24	OpenBMB	An end-side LLM outperforms Llama2-13B.		HuggingFace GitHub News

Multimodal Models

Nodel Name	Release	Developer	Description	Lang	Links
Qwen-VL-Max	06.02.24	Alibaba Cloud	This WebUI is based on Qwen-VL-Max, the upgraded version of Qwen-VL	CN	Homepage HuggingFace GitHub
Tiangong 2.0 MoE	06.02.24	Kunlun Wanwei	Supports multi-modal applications such as graphic and text dialogues, and text and text graphics, and supports ultra-long context windows of up to 100K. Probably some specific access is required for non China users.	CN	Homepage News
MiniCPM-V (i.e., OmniLMM-3B)	01.02.24	OpenBMB	It is an efficient version with promising performance for deployment. The model is built based on MiniCPM-2B and SigLip-400M, connected by a perceiver resampler.	CN EN	HuggingFace GitHub Demo

Text-to-Image

Model Name	Release	Developer	Description	Lang	Links
PixWeaver	30.04.24	Shengshu Technology	Image generation.	CN	HomePage
SDXL-Lightning	22.02.24	BiteDance	SDXL-Lightning is a lightning-fast text-to-image generation model. It can generate high-quality 1024px images in a few steps.	RU CN EN	arXiv HuggingFace Demo News

Text-to-Video

Model Name	Release	Developer	Description	Lang	Links
Vidu	29.04.24	Shengshu Technology	The firm has released several demo clips produced by Vidu, which can generate 1080p videos as long as 16 seconds	CN	HomePage News

Audio2Video

Model Name	Release	Developer	Description	Lang	Links
EMO	27.02.24	Alibaba	EMO, an expressive audio-driven portrait-video generation framework. Input a single reference image and the vocal audio, we can generate videos with any duration depending on the length of input video.	RU? CN EN	arXiv GitHub News

AI platforms

Platform Name	Release	Developer	Description	Lang	Links
Coze	01.02.24	BiteDance	Coze is a next-generation AI Bot development platform. Regardless of your programming experience, Coze enables you to effortlessly create various chatbots and deploy them across different social platforms and messaging apps like Discord, Telegram, LINE & Slack!	CN EN	Coze.com Coze.cn News

Voice Assistants

Voice Assistant	Release	Developer	Description	Lang	Links
Celia / Xiaoyi / AI Voice	2019	Huawei	Celia answers your questions and helps you get things done hands-free, so you can focus on the more important things in life.	CN EN RU? +4	Homepage Support News

Leaderboards

Leaderboard	Release	Developer	Description	Lang	Links
CompassRank		OpenCompass	It is a platform focused on evaluation of AGI, include Large Language Model and Multi-modality Model.	CN EN	HuggingFace GitHub News

You can find Chinese AI news on "Китайский Искусcтвенный Интеллект | 中国人工智能".