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 5, 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 32 stories · last 7 days
Newest first. Wider windows expand what is available; open Event History for the full period.
Today · Nov 5, 2026 6 stories
Reportedly: Baidu open-sources ERNIE-5.5-Flash-Base, a 280B-parameter MoE model with 14B active parameters
Baidu released and open-sourced ERNIE-5.5-Flash-Base, a sparse MoE model with 280B total and 14B active parameters, trained on 33.5T tokens. It supports 128K training context and …
Why it matters Baidu's open-weight release gives developers outside China direct access to a 280B-parameter MoE base model with 14B active parameters, reducing the cost of building on a state-of-the-art Chinese…
Reportedly: China Merchants' LionRock AI Lab Open-Sources Lion-ER-35B-A3B Embodied Reasoning Model
China Merchants' LionRock AI Lab open-sourced Lion-ER-35B-A3B, a 3B-active embodied reasoning model built with a capability-driven data engine that beat its base on ten benchmarks…
Why it matters Developers outside China gain access to a new open-weights embodied reasoning model, potentially reducing the cost and time to build robotics applications.
Reportedly: Tencent Hunyuan releases Hy4, joining the strongest open-source model ranks
Tencent Hunyuan released its latest large language model Hy4, which ranks among the strongest open-source models. The release was announced on November 5, 2026. No further details…
Why it matters Developers and enterprises evaluating open-source models gain a new candidate from Tencent, but without published benchmarks or weights, they cannot yet assess its cost or performance relative to…
Reportedly: USTC releases open-source RNA sequence foundation model structRFM, published in Nature
University of Science and Technology of China (USTC) released open-source RNA sequence foundation model structRFM, improving performance on RNA structure prediction and function i…
Why it matters Researchers and pharmaceutical companies working on RNA biology gain a new open-source tool that could reduce the cost and time of RNA structure prediction and function analysis, potentially…
Reportedly: Apple Intelligence launches in mainland China with Alibaba's Qwen integration
Apple Intelligence opened beta testing in mainland China on November 4 with iOS 27.3 Developer Beta, integrating Alibaba's Qwen into Siri, Writing Tools, and Visual Intelligence. …
Why it matters Apple's partnership with Alibaba gives Chinese iPhone users access to Apple Intelligence features powered by Qwen, potentially increasing iPhone competitiveness in China and expanding Alibaba's…
Reportedly: Huawei Ascend announces 0-day inference support for Moonshot AI's Kimi K3.1 Pro, a 2.8-trillion-parameter MoE model
Huawei Ascend announced 0-day inference support for Moonshot AI's Kimi K3.1 Pro, a 2.8-trillion-parameter MoE model with native vision and a million-token context window. The supp…
Why it matters Developers and enterprises using Huawei Ascend hardware can deploy Kimi K3.1 Pro on the day of its release, avoiding the usual delay for hardware-specific optimization.
Yesterday · Nov 4, 2026 2 stories
Reportedly: HongYun's HyperLM V2.5 Adds Native Support for Huawei Ascend 950DT and Atlas 850E SuperPoD
HongYun's HyperLM V2.5 now natively supports Huawei's Ascend 950DT and Atlas 850E SuperPoD, adding UnifiedBus 2.0 support, mxFP4 tuning, vNPU slicing, and prefill-decode split inf…
Why it matters Huawei Ascend hardware users gain access to HyperLM V2.5 with native optimizations, potentially reducing inference costs and improving performance on Ascend-based systems.
Reportedly: GPUWorks Open-Sources 40MB Inference Platform Running Ascend, PPU, Hygon and NVIDIA in One Cluster
GPUWorks open-sourced a 40MB AI inference platform under Apache-2.0 that pools Ascend, T-Head PPU, Hygon and NVIDIA GPUs in one cluster. Its developers say one server ran 1,003 GP…
Why it matters Developers outside China gain an Apache-2.0 licensed tool to unify Ascend, Hygon, PPU, and NVIDIA GPUs, reducing vendor lock-in and simplifying multi-accelerator cluster management.
Nov 3, 2026 2 stories
Reportedly: Ant Lingbo showcases fully open-source robot general brain at Wuzhen Summit
Ant Group's embodied AI unit Ant Lingbo presented its fully open-source robot general brain at the 2026 World Internet Conference Wuzhen Summit. LingBot series models have surpass…
Why it matters Robot developers outside China can now use Ant Lingbo's open-source models to avoid training robot brains from scratch for each robot configuration and task, reducing development cost and time.
Reportedly: Moonshot AI launches Kimi game development and creative work solution with direct Unity, Unreal Engine operation
Moonshot AI's Kimi released a game development and creative work solution based on Kimi K3.1 models, Kimi Code, Kimi Work agents, and plugins for Unity, Unreal Engine, and Blender…
Why it matters Game development teams using Unity, Unreal Engine, or Blender gain a tool that can operate their existing editors and projects, potentially reducing prototype iteration time and lowering the cost of…
Nov 2, 2026 10 stories
Reportedly: Mercedes-Benz launches 2026 long-wheelbase E-Class with ByteDance Doubao AI assistant, starting at RMB 423,800
Mercedes-Benz launched the 2026 long-wheelbase E-Class in China with a new AI cockpit integrating ByteDance's Doubao large model, IT Home reported. Prices start at RMB 423,800 (ap…
Why it matters Mercedes-Benz's adoption of ByteDance's Doubao model in its China-market E-Class gives ByteDance a foothold in premium automotive AI, potentially displacing Western AI providers in Chinese luxury…
Reportedly: XPENG MONA L05 Launches in China From RMB 139,800 With EV and Extended-Range Versions
XPENG launched the MONA L05 SUV in China at RMB 139,800-171,800 launch prices. It offers EV versions with up to 635 km range and a 1,310 km extended-range model, plus vision-only …
Why it matters Chinese consumers gain a new SUV option with extended-range capability and advanced driver assistance at a starting price of RMB 139,800, intensifying competition in the mid-size SUV segment.
Reportedly: DeepSeek raises 105.5 billion yuan in second funding round with investors including vivo, Meituan and Xiaohongshu
DeepSeek is raising 105.5 billion yuan in a second funding round, Leiphone reported, citing people familiar with the matter. Investors include vivo, Meituan and Xiaohongshu. The r…
Why it matters If confirmed, the reported funding would strengthen DeepSeek's capital position against competitors like Alibaba and Zhipu AI, potentially intensifying the Chinese large model price war and talent…
Reportedly: Tencent Cloud Launches WAND-Vega as Its WAND Platform Generates 4.5 Million Images a Day
Tencent Cloud launched WAND-Vega, a multimodal application model, alongside Real-time Intelligence and a contact-center voice agent. Its WAND platform serves over 2,000 firms, gen…
Why it matters Tencent Cloud's WAND platform now serves over 2,000 enterprises, meaning those businesses gain access to a production-scale multimodal generation service with daily outputs of 4.5 million images and…
Reportedly: Xiaomi open-sourced a 150-yuan AI gadget that can command agents and control smart home devices
Xiaomi open-sourced a 150-yuan AI gadget that can command agents and control smart home devices, QbitAI reported. The report did not specify the product name or release date. Acce…
Why it matters If confirmed, Xiaomi's open-source release would give developers outside China a low-cost reference design for agent-commanding smart home hardware, potentially lowering prototyping costs.
Reportedly: Mercedes-Benz launches 2026 E 300 L with China-developed AI cockpit, starting price cut by RMB 16,000
Mercedes-Benz launched the 2026 E 300 L sedan with a China-developed AI cockpit, IT Home reported. Starting price is RMB 423,800 (approx. $59,000), down RMB 16,000 from the previo…
Why it matters Mercedes-Benz is localizing its in-car AI for the Chinese market by integrating ByteDance's Doubao model, which could pressure other luxury automakers to adopt Chinese AI models to remain…
Tencent Hunyuan open-sources EmoReward, a self-evolving reward calibration mechanism for emotional video generation RL
Tencent Hunyuan released EmoReward, a self-evolving training mechanism that calibrates multi-reward weight allocation for emotional video-generation RL without updating reward mod…
Why it matters Developers working on emotional video generation can now use EmoReward to improve RL training stability and output quality without retraining reward models, potentially reducing compute costs and…
Moonshot AI released Kimi-K3.1-Pro on Hugging Face
Moonshot AI released Kimi-K3.1-Pro, an open-weight multimodal agentic model with 2.8T total parameters and a 1M-token context window, on Hugging Face. The model is post-trained fr…
Why it matters Developers outside China can now access a 2.8T-parameter open-weight multimodal model from Moonshot AI, but the custom license and lack of independent benchmarks mean adoption requires legal review…
Reportedly: Qwen AI Glasses Add Call Translation and Media Translation in New OTA Update
Qwen AI Glasses received an OTA update adding real-time call translation for phone calls and meetings, media translation shown in the companion app, and improved speech recognitio…
Why it matters Users of Qwen AI Glasses gain real-time call and media translation, reducing the need for separate translation devices or apps during conversations and media consumption.
Reportedly: UBTECH Robotics releases Thinker-Sim embodied AI simulation platform for robot capability evaluation
UBTECH Robotics released Thinker-Sim, an embodied AI simulation platform providing a one-stop system for real-scene digitization, large-scale simulation data synthesis, and robot …
Why it matters Robot developers using UBTECH's platform can reduce the cost and time of building simulation environments for training and evaluation, potentially accelerating embodied AI development.
Nov 1, 2026 0 stories
Reportedly: Pika open-sources H3 Flex, a MiniMax H3 derivative for joint audio-video generation in six steps
Pika released H3 Flex, a MiniMax H3 derivative with open adapter weights and inference code, Leiphone reported. It generates 480P/768P video with synchronized stereo audio in abou…
Why it matters Pika's open-weight H3 Flex gives developers outside China a deployable MiniMax H3 derivative for joint audio-video generation, reducing reliance on Pika's paid API for high-resolution output.
Oct 31, 2026 0 stories
Reportedly: DeepSeek Harness open-sources Super-agent architecture prototype
DeepSeek Harness released the dsh-superagent-prototype on GitHub, a Super-agent architecture that keeps context across sessions and runs multiple tasks in parallel. The prototype …
Why it matters Developers building persistent personal assistants get a reference architecture that removes the need to manually switch sessions and maintain context, potentially reducing the human attention…
Reportedly: Xiaomi open-sources 0.8B Xiaomi-OCR-0 document understanding model
Xiaomi released and open-sourced Xiaomi-OCR-0, a 0.8B-parameter document understanding model combining document parsing, key information extraction, and visual question answering.…
Why it matters Xiaomi's release of a sub-1B document understanding model that outperforms larger rivals on OmniDocBench v1.6 puts pressure on other mobile AI providers to match on-device OCR and document parsing…
Reportedly: Apple China launches Apple Intelligence intro page with first core features coming later this year, integrating Qwen
Apple's China website launched an Apple Intelligence introduction page, stating first core features will arrive later this year. IT Home reported the features will be available in…
Why it matters Apple's partnership with Alibaba gives Chinese iPhone users access to Apple Intelligence features, potentially reducing the competitive gap with local Android manufacturers that already offer AI…
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.