China AI news, from the source
The model, price, policy and company changes that affect what you build outside China — verified against primary Chinese sources.
Save Events across devicesFollow companies and labsChina AI brief in your inbox
Top stories
Updated Nov 20, 2026 Share this briefingToday's brief is being compiled. Verified Events keep landing in the live update ledger below and in the full Event History.
Browse the Event HistoryMore China AI news
Evidence-qualified product, pricing, policy, capital and deployment Events, ordered by when they happened. Research appears only after an impact review; unverified Signals stay separate.
20 of 28 stories · last 7 days
Newest first. Wider windows expand what is available; open Event History for the full period.
Today · Nov 20, 2026 5 stories
Reportedly: OpenDesign team's OpenMotion lets DeepSeek Harness create motion graphics
The OpenDesign team built OpenMotion, a plugin that Zhidongxi reported using with DeepSeek Harness to create a DSH-Go promotional video. The outlet said OpenDesign had more than 1…
Why it matters Developers using DeepSeek Harness can draft product demos through the agent, as Zhidongxi's DSH-Go test showed.
Reportedly: Xiaomi adds MiMo UltraSpeed to Token Plan Pro and Max for a limited time
Xiaomi is giving Token Plan Pro and Max subscribers limited-time access to MiMo-V2.6-Pro-UltraSpeed starting November 19, IT Home reported, citing Xiaomi’s MiMo documentation. The…
Why it matters If the reported access is available on their accounts, Token Plan Pro and Max subscribers can test a faster inference option without switching tiers.
MiniMax publishes cryptanalysis findings from its internal AI model
MiniMax published cryptanalysis findings that it says were primarily discovered by an internal AI model, covering encryption, digital signatures, and hash functions. The GitHub re…
Why it matters Cryptography maintainers outside China now have public MiniMax findings to check against the versions and configurations they support.
MiniMax publishes Cyber repository on GitHub
MiniMax published Cyber, a new open-source repository on GitHub. The source release was published on 2026-11-20.
Why it matters Developers can inspect Cyber on GitHub, but the listing alone does not establish what they can run or reuse.
Reportedly: Tencent opens Hy4 weights, completing the Hunyuan LLM open-weight rollout
Tencent made Hy4's weights available for download and self-hosting, completing the Hunyuan LLM series' open-weight rollout, IT Home reported. Hy4 reportedly has 770B total paramet…
Why it matters If the reported downloads and license are verified, developers can modify and self-host Hy4 for commercial use without relying on a Tencent-hosted API.
Yesterday · Nov 19, 2026 6 stories
Reportedly: Tencent sharply cuts investment in its Yuanbao AI assistant
Tencent is sharply cutting investment in its Yuanbao AI assistant after the product fell short of expectations, IT Home reported. The report did not quantify the cut.
Why it matters If the reported cut is confirmed, Tencent's Yuanbao team will have less investment behind the assistant.
inclusionAI releases a streaming guardrail for DeepSeek-V4.1-Flash-1017
inclusionAI released DeepSeek-V4.1-Flash-1017-singprobe, a 10.10M-parameter guardrail that scores query intent, response unsafety, and hallucination risk at each token. It reads h…
Why it matters Teams serving DeepSeek-V4.1-Flash-1017 can use the released probe to score response unsafety and hallucination risk before generation finishes.
Reportedly: Alibaba’s recurring revenue from external AI cloud and computing customers rose 67%
Alibaba’s recurring revenue from external AI cloud and computing customers rose 67% in the quarter ended September 30, SCMP Tech reported. Total revenue rose 10% to a better-than-…
Why it matters For Alibaba investors, the reported increase identifies recurring sales of AI cloud and computing services to external customers as a contributor to revenue growth.
Reportedly: Moonshot AI investigates degraded API availability
Moonshot AI is investigating degraded API availability that began at 17:25 on November 19, IT Home reported, citing the company's status dashboard. IT Home also reported a partial…
Why it matters Developers who rely on Moonshot AI's API may find that workflows depending on it are less reliable during the reported incident.
Reportedly: Anthropic prices $81.3 billion IPO at $83 a share
Anthropic priced its $81.3 billion IPO at $83 a share, SCMP Tech reported. The price exceeded the marketed range of $70 to $78. It implied a market capitalization of about $1.5 tr…
Why it matters Lower-cost Chinese model providers would face a better-funded rival if Anthropic's reported $81.3 billion IPO is confirmed.
SenseTime launches Raccoon Work for finance
SenseTime launched Raccoon Work for AI applications in finance.
Why it matters Finance teams have a new SenseTime product to evaluate, but the available evidence does not establish whether they can access it or use it in their existing workflows.
Nov 18, 2026 4 stories
Reportedly: Tencent expects better returns on AI infrastructure spending by 2028
Tencent executives expect AI infrastructure spending to generate more value by 2028 through infrastructure reuse, lower hardware prices and more efficient compute, IT Home reporte…
Why it matters For investors assessing Tencent's cash needs, the reported compute prepayments turn a positive third-quarter 2026 free cash flow figure excluding those payments into a negative one when they are…
Reportedly: Doubao Work tops 8 million daily active users within three months of launch
Doubao Work said it surpassed 8 million daily active users and RMB 2 billion in enterprise annual recurring revenue within three months of launch, IT Home reported. IT Home also r…
Why it matters Enterprise AI vendors benchmarking their China business have a new, company-reported revenue comparison point: more than RMB 2 billion in enterprise annual recurring revenue for Doubao Work.
Reportedly: Beijing Academy of Artificial Intelligence and Qingmao Intelligence launch commercial FlagOS token factory on Chinese-made chips
The Beijing Academy of Artificial Intelligence and Qingmao Intelligence launched a commercial FlagOS token factory on Chinese-made chips, Leiphone reported. The partners unveiled …
Why it matters For operators of Chinese-made accelerator clusters, the reported deployment provides a commercial example of using a shared inference stack instead of maintaining a separate stack for each chip type.
SenseTime reports faster revenue growth for the first nine months of 2026
SenseTime reported that revenue rose 30.4% year over year to RMB 3.79 billion in the nine months ended September 30, 2026, with growth accelerating from the first half of 2026.
Why it matters Investors tracking SenseTime now have a nine-month revenue figure to compare with the company's full-year results.
Nov 17, 2026 4 stories
Tencent Hunyuan publishes the Hy4 Flash repository and links to GGUF weights
Tencent Hunyuan published the Hy4 Flash repository, documenting a model with 220B total parameters, about 12B active per token, and a context window of up to 1M tokens. It accepts…
Why it matters Developers outside China can access Hy4 Flash through the linked GGUF weights without relying on a Tencent-hosted API.
Unitree Robotics open-sources UNICALIB for robot camera calibration
Unitree Robotics open-sourced UNICALIB 0.1.7, a browser-based workbench that calibrates robot cameras and automatically checks whether the results pass. It supports intrinsic, ext…
Why it matters Unitree robot integrators can use one browser interface to capture calibration images, check the results and write them back to the robot.
Reportedly: Apple adds AIGC compliance fields to China App Store Connect: existing apps must complete filing by Jan 31, 2027
Apple added two mandatory compliance fields to App Store Connect for apps distributed in mainland China: an AIGC service declaration and filing information, and proof of completed…
Why it matters App developers distributing AIGC-enabled apps in China's App Store now face a hard compliance deadline of January 31, 2027, requiring them to obtain algorithm filings and security assessments or…
Reportedly: Huling Technology develops an AI voice ring after seed funding
Huling Technology is developing an AI voice ring and has raised tens of millions of yuan in seed funding, IT Home reported, citing people familiar with the matter. The ring is rep…
Why it matters If the reported product reaches buyers, voice recording and task capture could become available through a ring rather than a screen-based device.
Nov 16, 2026 1 story
Reportedly: Baidu reports Q3 2026 revenue of 30.8 billion yuan, with AI business contributing 53% of core revenue
Baidu reported Q3 2026 revenue of 30.8 billion yuan, down 1% year-over-year, with AI business revenue reaching 13.2 billion yuan, 53% of core revenue. AI cloud infrastructure reve…
Why it matters Baidu's AI cloud infrastructure revenue growth of 94% year-over-year shows that Chinese enterprises are increasingly paying for AI compute and model services, which pressures competitors like…
Reportedly: Tencent releases Hy4 flash: 224B total parameters, 12B active, per-task cost about 5% of Hy4
Tencent's Hunyuan team released Hy4 flash, a 224B-parameter model with 12B active per token, 1M-token context, and multimodal input. API pricing is much lower than Hy4; on WorkInd…
Why it matters Developers using Tencent Hunyuan APIs can now get Hy4-level intelligence at about 5% of the per-task cost, directly undercutting DeepSeek's flash-tier pricing by 8x on the cited benchmark.
Reportedly: Qianwen Office adds 10 models including Qwen4-Max, Kimi-K3.1, GLM-5.3
Qianwen Office added 10 models on Nov 16, including Qwen4-Max, Kimi-K3.1, GLM-5.3, DeepSeek, MiniMax, and Step. Personal and enterprise standard users can now choose models freely…
Why it matters Qianwen Office users gain lower token costs and better task fit by choosing among 10 models, while enterprise admins get granular control over model access.
Reportedly: Huawei Kirin 9050 series chip details revealed: 30% audio performance boost, first H.266 hard decoding, low-latency AI heterogeneous computing
Huawei revealed new details of its Kirin 9050 series chip, including a 30% audio performance boost, first H.266 hard decoding, and low-latency AI heterogeneous computing. The chip…
Why it matters Huawei's Kirin 9050 series with H.266 hard decoding support could pressure other chipmakers to adopt the newer codec, potentially reducing bandwidth costs for streaming services and improving user…
Reportedly: Alibaba releases Qwen4-Max, a 2.4-trillion-parameter MoE flagship model with 1M-token context and over 60% lower task cost
Alibaba's Qwen team released Qwen4-Max, a 2.4-trillion-parameter sparse MoE flagship model with native vision and a 1M-token context window. It is available via the Qwen AI platfo…
Why it matters Developers using Qwen's API face lower per-task costs and a longer context window, which may reduce the need for external retrieval or summarization pipelines.
No additional verified events in this window. Try a wider period or open Event History.
View All eventsNew Chinese AI models
All models- MeituanLongCat-2.5Nov 1, 2026
- Moonshot AIKimi K3.1 ProOct 23, 2026 · Global API
- Moonshot AIKimi K3.1 Pro HighspeedOct 23, 2026 · Global API
- MiniMaxMiniMax-M3.1-FlashOct 19, 2026 · Open weights · Global API
- DeepSeekDeepSeek-V4.1-Flash-1017Oct 17, 2026 · Open weights · Global API
- AlibabaQwen-Image-2.1-TurboOct 9, 2026 · Open weights
Explore China AI
- China AI modelsAccess, open weights, and licenses across Chinese labs.
- Chinese AI companies73 frontier labs, big tech, robotics and AI chip makers.
- China AI policyRules, algorithm filings, labeling standards and export controls.
- Attention IndexEnglish and Chinese source attention.
- China AI ExplainersEssential primers on terminology, frontier startups, and compliance.
- China AI LinesFive strategic themes: open weights, price wars, robotics, chips and policy.
Your reading list
Save stories. Free, across devices.