China AI

AIArtificial Analysis says Bria's Fibo 2.0, an 8B parameter open weights image model, ranks first among open weights models on AA-Image-Editing v2.1 and is statistically tied with Qwen-Image-2.1 on AA-Image-T2I v2.1.

AIAn anonymous account tracking AI talent moves reported on Tuesday that Nathan Chen Guangyu, 17, is joining Discovery Loop, a two-month-old AI startup founded by four former Google senior researchers including Jeff Dean.

AIRohan Paul shares results for an RL post-trained Qwen3.8-27B model trained with open-source environments.

AIOpenBMB says MiniCPM-V 4.7, a 35B multimodal model, averages 20.51 tokens per second on an iPhone 18 Pro Max with 12GB of memory.

AIThe openbmb/tiny-processor-minicpm-ocr repository on Hugging Face is publicly listed, but users must accept its conditions and share their Hugging Face information before viewing its files.

AINeolix, an autonomous delivery vehicle company, says it has closed a Series E round of nearly $400 million led by Boyu Capital and a national-level fund.

AISpacemiT and the XiangShan team introduced the V100 RISC-V server CPU alongside three complete servers at the RISC-V China Summit on October 19.

AIWuXi Stars Micro System Technologies introduced three interconnect chips for AI servers and clusters at its first product launch on October 15.

AIPhysis and vLLM introduce PhysisServe, a serving system that uses stage-wise pipelining, time-sharing, and joint multi-stream scheduling to improve both single-stream latency and multi-stream efficiency.

AIPerch AI now offers MiniMax M3.1 Flash free of charge, alongside GLM 5.3 Flash, DeepSeek V4.1 Flash, and Beam 1.0.

AIZhongke Tongliang showed its SmarCo AI workstation, SmarCo intelligent video encoding appliance and SmarCo AI acceleration card at the 6th RISC-V China Summit in Shenzhen on October 18.

AIVideoJev-4B, a Qwen3.5-4B-based model from MCG-NJU, scores candidate answers and time spans directly for video question answering and temporal grounding.

AIAnt Group releases Ming-3.0-tts, an open-weights text-to-speech model under the MIT License, with a 0.5B dense version and a 7.9B mixture-of-experts variant named Ming-3.0-tts-tiny.

AIOllama says MiniMax M3.1 Flash is free on Ollama Cloud for the next three days, in collaboration with MiniMax.

AISGLang announced a Low-Latency Mode built with NVIDIA and AMD that raises P90 per-user token speed on DeepSeek-V4.1-Flash.

AIDesign Arena projects DeepSeek-V4.1-Flash-1017 to rank #3 among open-weight models on its Website leaderboard, a 16-position jump past DeepSeek-V4-Flash-0731.

AIModelScope now hosts the MiniMax H3 Character Swap LoRA, an experimental adapter that recasts a selected person in video from an image reference.

AIHeyGen Video 1.0, a 1080p model post-trained on MiniMax H3, ranks #3 on both Design Arena's Video and Image to Video leaderboards, according to Design Arena.

AIOpenCode says its DeepSeek V4.1 Flash model has been updated to 1017 in OpenCode Go and the Free tier. It adds that no changes are needed on users' end.

AIZhongxingwei, a Chinese interconnect chip company, unveiled 12 mass-produced chips and three new high-performance interconnect products at a launch event.

AIErik Kannike releases qwen3-4b-deslop, a Qwen3-4B-Instruct fine-tune trained on a MacBook Air to avoid em dashes, "not X, but Y" constructions and lists of three without a system prompt.

AIInferact says it is an API launch partner for MiniMax M3.1 Flash and will provide the production inference serving stack using its vLLM engineering and Inferact Platform.

AILightmake, a 3D printing company founded in April 2025 by Wang Zhiyu, says it is focusing on small businesses and print farms rather than consumers.

AIModelScope says the MiniMax H3 Character Swap LoRA is now available, an experimental adapter that recasts a selected person in video using an image reference.

AIMemSilicon, a Chinese chip company, has closed a Series A round of over 400 million yuan led by Boyuan Capital.

AIMemSilicon has closed a Series A round of over 400 million yuan, led by Boyuan Capital, to fund RRAM commercialization and development of its LPU, an inference chip for large models.

AIRokid opened an unstaffed flagship smart-glasses store in Hong Kong's Harbour City, where customers test devices, pay on screens and collect products from a robotic system without any sales staff.

AIGlenfly Tech says its first Arm-based system-on-chip, the GF800, has finished chip debugging and is sampling to end customers.

AIvLLM reports Kimi K3 decode reaching 842 tokens/s at batch 1 on NVIDIA GB300 NVL72 using DSpark.

Why it matters: The post gives concrete throughput, interactivity and configuration results, showing how a unified decode kernel changes serving performance for a large model.

AIOllama says DeepSeek-V4.1-Flash-1017 is available on Ollama Cloud, hosted in the US with zero data retention.

AIZevio says DeepSeek V4.1 Flash 1017 is now available on its platform at 45% lower inference cost.

AIBolt is adding GLM 5.3 Flash to every account's model picker this weekend at no cost, with no plan change or code required. The post is promoted as part of Bolt Builders Week, which offers a $10k prize for apps submitted by October 20.

AIvLLM congratulates DeepSeek on DeepSeek-V4.1-Flash-1017, which the post says has better scores and shorter outputs for long agent loops. vLLM says it delivers 1.3B tokens per dollar TCO on SemiAnalysis AgentX, at 830K tok/s per GPU (cached + input + output) at 100 tok/s per user.

AIPurple Mountain Laboratories and China Mobile have launched AXP, an agent interconnection platform built on deterministic networking.

AIShenchen Kaidong's badminton robot completed 1,452 consecutive rallies with a human player on December 22, 2025, earning a Guinness World Record for most consecutive counter hits by a mobile robot.

AIWu Haixu, first author of the GeoPT paper, has founded Wuyuan Technology, a world-model startup that has raised several rounds and reached a valuation of several hundred million US dollars, with investors including Ant Group and Meituan.

AIRokid has launched ride-hailing on its AI glasses with Tencent Mobility, letting users book a car by voice and pay without a phone.

AIDesign Arena announces that Ultramarine 1 Flash, from @paperinstr, is now available on its platform.

AISemiAnalysis says Amazon blocked Meta's Muse after it reached #1 on the App Store, and Jefferies found a third of tested US shopping and travel sites also block it.

AIvLLM says eight AMD MI355X GPUs deliver 24.4% more throughput than an HGX B300 on Kimi K3 serving, at 7,567 versus 6,083 tokens per second per GPU at 50 tokens per second per user.