Chinese AI
Companies, apps, open models, daily news and physical AI products from China — with official links.
Chinese AI models, apps and companies, from DeepSeek and Qwen to Kimi and GLM.
Companies & apps
- DeepSeek
Hangzhou
DeepSeek-V3, R1 (open weights)
- Alibaba Qwen
Hangzhou
Qwen3 family, Qwen-VL, Qwen-Image
- Moonshot AI (Kimi)
Beijing
Kimi K2 (open weights)
- Zhipu AI / Z.ai (GLM)
Beijing
GLM-4.5, GLM-4.6, CogVideoX
- Baidu (ERNIE)
Beijing
ERNIE 4.5 / X1, Apollo Go
- ByteDance (Doubao / Seed)
Beijing
Doubao, Seed-OSS, Seedream, Seedance
- Tencent (Hunyuan)
Shenzhen
Hunyuan-Large, HunyuanVideo, Hunyuan3D
- MiniMax
Shanghai
MiniMax-M1/M2, Hailuo video
- StepFun
Shanghai
Step-2, Step-Video
- 01.AI
Beijing
Yi series
- Kuaishou (Kling)
Beijing
Kling video generation
- Shanghai AI Lab (InternLM)
Shanghai
InternLM, InternVL
- iFlytek (Spark)
Hefei
Spark LLM, speech AI
- SenseTime (SenseNova)
Shanghai
SenseNova
- OpenBMB / ModelBest (MiniCPM)
Beijing
MiniCPM on-device models
Daily Chinese AI updates
- Hugging Face: Thinking Machines Lab releases Inkling 2 Mini, with AI layers on repeatDigg · 22 Nov 2026
- Cotton accuses Lutnick of ignoring alerts on AI chip smuggling to ChinaNewsNation · 22 Nov 2026
- Thinking Machines wows with Inkling 2 Mini looped transformer model that runs locally on laptopsVentureBeat (main) · 21 Nov 2026
- CertiK and Columbia University Researchers Formally Verify Production AI Kernels From DeepSeek and FlashAttentionCoinDesk · 21 Nov 2026
- Bessent says no further restrictions are needed for Nvidia chip exports to ChinaPBS · 21 Nov 2026
- China’s military sought DeepSeek for cyber drillsSocial News XYZ · 21 Nov 2026
- Meet Thinking Machines Inkling 2 Mini: 26B Looped MoE Runs on a 32GB MacMarkTechPost · 21 Nov 2026
- Amazon: China-Linked Hackers’ AI Agents Exploited Medical Software Zero-DaySecurityWeek · 21 Nov 2026
- US Treasury Secretary Says Nvidia Chip Export Policy to China Holds After Remote-Access CurbsCASRAI · 21 Nov 2026
- Kyle Chan: Why Beijing Won't Stimulate Consumers — and What That Means for Huawei, DeepSeek and the Chip RaceBigGo Finance · 21 Nov 2026
- New open model Inkling 2 Mini thinks in loops to solve problems with fewer words and still runs on a laptopThe Decoder · 21 Nov 2026
- TSMC Sinks 20% in a Week on Nvidia Jitters and New US Chip Curbs on China. Time to Buy?TIKR.com · 21 Nov 2026
- Bytedance argues most AI training phases are unnecessary and backs it up with a competitive open modelThe Decoder · 21 Nov 2026
- Nvidia’s latest chip scandal has an Alibaba connection, new report claimsYahoo Finance · 21 Nov 2026
- Trump Announces US-China AI Safety Deal After Meeting XiCryptoRank · 21 Nov 2026
- Sarvam AI's Token Factory offers Chinese AI models on Indian servers: What inference, open-weight models and the sovereignty debate mean for India's AI missionEducation Times · 21 Nov 2026
- India, China, Nigeria lead countries using AI at workBusiness News Nigeria · 21 Nov 2026
- DeepSeek’s First Founder: The Public Unveiling of the AI Lab’s Core Leadership36Kr · 21 Nov 2026
- CertiK and Columbia University Researchers Release TensorLean Results, Completing the World's First Formal Verification of Production-Grade AI Operators in DeepSeek and FlashAttentionPANews · 21 Nov 2026
- DaisyRose 27B vs Qwen 3.8 27B: What Pure RL ChangedOrcaRouter · 21 Nov 2026
- Grimes Shares Two New Singles From PSYOPERA “Killing God With Computers” & “Deepseek”mxdwn Music · 21 Nov 2026
- Inkling 2 Mini vs Qwen 3.8: the Tier That CompetesOrcaRouter · 21 Nov 2026
- Anysphere Weighs Moving Cursor Away From Chinese Open-Weight ModelsUnite.AI · 21 Nov 2026
- Treasury secretary says no further restrictions are needed for Nvidia chip exports to ChinaNews4JAX · 21 Nov 2026
- US lawmakers push for tougher AI chip rules amid China export crackdownAl Jazeera · 21 Nov 2026
- Trump’s Treasury Secretary Keeps Nvidia’s H200 China Policy as House Data Show Nearly 880,000 AI Chips ApprovedYahoo Finance · 21 Nov 2026
- Officials approved the sale of 580,000 Nvidia chips to Chinese companiesThe Washington Post · 21 Nov 2026
- Thinking Machines Lab Releases Inkling 2 Mini for Local DevicesUnite.AI · 21 Nov 2026
- Microsoft slides while warning US ban on Chinese transceivers may disrupt AI infrastructure expansionTraders Union · 21 Nov 2026
- Nvidia Stock Plunges 12% After Reuters Reports an SEC Probe of Its Accounting. It's Nvidia's Worst Day Since DeepSeek and It Has Lost $2.7 Trillion Since October24/7 Wall St. · 21 Nov 2026
Top Chinese open models
- Qwen/Qwen3.8-27B♥ 20774
- deepseek-ai/DeepSeek-R1♥ 14443
- moonshotai/Kimi-K3♥ 11962
- Qwen/Qwen3.8-Flash-Next♥ 7304
- MiniMaxAI/MiniMax-H3♥ 7174
- deepseek-ai/DeepSeek-V4.1-Flash-1017♥ 5780
- deepseek-ai/DeepSeek-V4-Pro♥ 5700
- zai-org/GLM-5.2♥ 5206
- baidu/Unlimited-OCR♥ 4888
- deepseek-ai/DeepSeek-V4.1-Flash♥ 4724
- deepseek-ai/DeepSeek-V3♥ 4462
- deepseek-ai/DeepSeek-V4-Flash-0731♥ 4099
- deepseek-ai/Janus-Pro-7B♥ 3722
- Qwen/Qwen-Image-2.1♥ 3697
- moonshotai/Kimi-K3.1-Pro♥ 2305
- MiniMaxAI/MiniMax-H3-Image♥ 1664
- openbmb/MiniCPM-V-4.7-35B-A3B♥ 1071
- Qwen/Qwen-Image-2.1-Turbo♥ 1034
- baidu/Unlimited-OCR-2♥ 668
- MiniMaxAI/MiniMax-M3.1-Flash♥ 537
- tencent/Hy4-flash♥ 512
- tencent/Hy-SimulMT-30B-A3B-FP8♥ 458
- openbmb/MiniCPM-OCR-1.0♥ 441
- openbmb/MiniCPM-V-4.7-1B♥ 394
- Qwen3-0.6B30,452,693
- Qwen3-4B12,585,505
- Qwen3-8B9,739,109
- Qwen3-Embedding-0.6B8,598,074
- Qwen3.5-9B8,489,640
- Qwen3.5-4B7,946,136
- Qwen3-VL-8B-Instruct7,390,307
- Qwen2.5-7B-Instruct7,357,696
- Qwen3.8-27B7,221,327
- Qwen3-14B6,913,178
- GLM-5.3-Flash6,515,511
- Qwen2.5-0.5B-Instruct6,276,110
- Qwen2.5-1.5B-Instruct6,018,399
- Qwen2.5-VL-7B-Instruct5,132,099
- DeepSeek-V3.24,844,451
- Qwen3.8-27B-FP84,470,739
- DeepSeek-V4-Flash-07314,438,887
- Qwen2.5-3B-Instruct4,299,162
- Qwen3-1.7B4,034,338
- Qwen3.6-35B-A3B-FP84,021,372
- MiniMax-H33,765,471
- Qwen3.6-35B-A3B3,667,555
- Qwen3-4B-Instruct-25073,293,032
- Qwen3.5-2B3,243,646
- Qwen3-32B3,235,453
- Qwen3-VL-4B-Instruct3,208,816
- Qwen3-TTS-12Hz-1.7B-Base3,162,646
- Qwen3-14B-AWQ2,894,022
- Qwen3.5-0.8B2,820,039
- Qwen3-VL-2B-Instruct2,385,218
Chinese physical AI
- Unitree G1
Unitree (China)
Humanoid, ~1.3 m, 23–43 DoF
Official ↗ - Unitree Go2
Unitree (China)
Quadruped robot dog, LiDAR, GPT-enabled voice
Official ↗ - UBTech Walker S
UBTech (China)
Industrial humanoid for factories
Official ↗ - DJI Mavic 4 Pro
DJI (China)
Hasselblad camera, omnidirectional obstacle sensing
Official ↗ - DJI Neo
DJI (China)
135 g palm-launch AI subject tracking
Official ↗ - Baidu Apollo Go
Baidu (China)
Driverless robotaxi, RT6 vehicle
Official ↗ - Pony.ai
Pony.ai (China)
Robotaxi & robotruck, Gen-7 system
Official ↗ - WeRide
WeRide (China)
Robotaxi, robobus, robosweeper
Official ↗ - Xiaomi 15 / HyperAI
Xiaomi (China)
HyperOS AI features
Official ↗
Frequently asked questions
Are Chinese AI models open source?
Several are released with open weights; each listing shows its license.
Related
Comments
No comments yet. Share your own experience below.
Visitor posts are moderated before they appear and are the opinion of their authors, not DEEKOYA AI. Names are not verified. See our privacy policy.