All news

LLatent Space12:47Model releases

AI score83

Google DeepMind releases Gemini 4 Carbon, topping Artificial…

Google DeepMind has released Gemini 4 Carbon, which ranks first on the Artificial Analysis Intelligence Index v5 with a score of 59 and leads in 4 of 9 evaluations including HLE-Diamond and TB-Science. The model focuses on cybersecurity and long-horizon tasks, claiming 88.2% on SEC-bench Pro, 54.8% on ExploitGym, and 71.0% on CWE-bench, and is initially available only to trusted cybersecurity personnel through the Fairwind Program. In terms of specifications, it supports a 1M token context and 1M maximum output, with input covering text, images, video, and speech, and output in text; Artificial Analysis lists a standard price of $4/$20 per million input/output tokens, plus a 50% launch discount.

Same story · 34 reports

#Model releases#Google#Safety / alignment

LLatent Space12:47Analysis

AI score54

Elorian Releases DecoyCount Benchmark: Frontier Models All Score Below 50% on Visual Counting Accuracy

Elorian and Mercor jointly released the DecoyCount benchmark, which tests whether models truly count objects or rely on visual shortcuts. Frontier models all scored below 50% accuracy, with the highest being GPT-6 Astra at 43.17%. Elorian says its visual reasoning model Glimpse scored 50.48% in a single run on 1,660 questions, making it the only model to exceed 50%, but this result comes from internal testing and did not use an agent or tool framework, whereas the six models in the paper each took the average of three runs. Elorian CEO Andrew Dai said the benchmark examines both vision and reasoning, and can distinguish whether a model failed to see the objects or miscounted them.

#Evaluations / benchmarks#Multimodal#Reasoning

CChina.com Technology12:47

AI score28

DingTalk A1 Doukou Doctor Edition: Connecting Voice Records, AI Summaries, and Medical Evidence

DingTalk A1 Doukou Doctor Edition combines A1's voice recording and AI summary capabilities with Doukou's medical evidence capabilities. After doctors finish organizing case discussions or lecture content, they can also search literature, guidelines, and expert consensus around unresolved medical questions. A gynecology department director at a tertiary hospital in Hangzhou reported that on surgery days she can first record special situations and judgments, then organize them after surgery, while for complex outpatient visits she uses Doukou Doctor to view literature and guideline links. At this stage, it is more like a portable professional assistant that can record, distill, and search literature, rather than a reconstruction of hospital IT systems.

#Product updates#Speech#RAG

CChina.com Technology12:47

AI score27

Lingxie AI Explores Personal Knowledge Services: Building Conversational Digital Avatars with Personal Knowledge Bases

Lingxie AI, based on personal knowledge bases and instant messaging, connects creators' accumulated content with user questions to build a conversational digital avatar for everyone. Public version records show that version 1.0.7 already includes avatar subscriptions, paid groups, and creator revenue features, and supports knowledge base website imports and chat favorites bound to avatars to participate in answering. A CNNIC report shows that as of the first half of 2026, China's generative AI user base exceeded 700 million, with a penetration rate of over 50%, of which 76.0% of users use such tools to answer questions.

#Product updates#Agents

CChina.com Technology12:47

AI score14

Yadea Showcases Guineng White Shark II at Yunqi Conference: Ten-Thousand-Yuan-Class Smart Experience at a Four-Thousand-Yuan Price Point

At the 2026 Yunqi Conference, Yadea exhibited the Guineng White Shark II Ellen Edition, offering a ten-thousand-yuan-class smart experience at a four-thousand-yuan price point, equipped with the new-generation Yadea OS, the Smart display, the smart travel assistant "Xiaodi," and iRide keyless vehicle control. The vehicle is a collaboration with "Zenless Zone Zero," featuring an exclusive UI and the Fairy smart system. Yadea also jointly established the "Two-Wheeler Carbon Inclusion Alliance" with the Amap Open Platform, becoming one of the first batch of ecosystem partners.

#Industry news#Embodied AI

CChina.com Technology12:47

AI score58

AI demand drives chip business, Samsung operating profit surges 783%

Samsung Electronics announced its financial results for the third quarter of 2026, with revenue of approximately 195.3 trillion won, up about 127% year-on-year, and operating profit of approximately 107.4 trillion won, up about 783% year-on-year, marking the fourth consecutive quarter of record highs. The semiconductor division's operating profit for the quarter reached 106.5 trillion won, accounting for about 99% of the company's total, with HBM4 sales rising sharply and demand for server SSDs strong. Revenue from the mobile and network businesses grew but they still posted a combined loss of 600 billion won, which Samsung attributed to cost pressure from rising component prices.

Same story · 3 reports

#Industry news#English text is being prepared.

CChina.com Technology12:47Model releases

AI score77

Google Releases Gemini 4 Carbon Large Model, with a Focus on Enhancing Cybersecurity Capabilities

Google has launched its new-generation large model Gemini 4 Carbon, focusing on cybersecurity and long-horizon complex task handling, and it ranks first with an overall score of 59 on the Artificial Analysis Intelligence Index v5.0. On WebDev Arena, the model improved by 89 Elo points over the previous generation to 1769 points, ranking fourth; internal testing shows that its success rate in outputting secure code is more than twice that of Gemini 4 Argon. Pricing is initially the same as Gemini 4 Argon: during the promotional period, input is $2 per million tokens and output is $10 per million tokens. This week it is being made available for trial to a small number of cybersecurity practitioners on a priority basis.

Same story · 34 reports

#Model releases#Safety / alignment#Agents

AAnt inclusionAI: New GitHub Repository12:47First-party

AI score43

Ant inclusionAI Open-Sources the UBTree Project Page: Using Unigram and…

Ant inclusionAI has launched the UBTree project page on GitHub, introducing a speculative decoding method that uses Unigram and Bigram models for parallel tree drafting (arXiv:2609.39972). The page includes demo videos of Ling3 Tiny and Flash, comparing four methods—AR, DFlash, DSpark, and UBTree—under the same target model; the videos are 1440×760, 20 FPS, at 0.25× playback speed. The figures and table numbers are all driven by static/js/data.js.

#English text is being prepared.#Reasoning#arXiv

AAnt inclusionAI: New GitHub repository12:47First-party

AI score35

Ant inclusionAI Open-Sources the ZwZ Model Family and the ZoomBench Fine-Grained Perception Benchmark (ICML 2026)

Ant inclusionAI has released the ZwZ model family (4/7/8B) accompanying its ICML 2026 paper, achieving SOTA on multimodal perception benchmarks among open-source models, while requiring only a single forward inference pass and no tool calls at inference time. The method is Region-to-Image Distillation (R2I), which distills high-quality VQA supervision generated from enlarged local regions back into the full image, with 37K training samples. It also introduces ZoomBench, containing 845 VQA items covering 6 fine-grained perception dimensions, on which Qwen2.5-VL-7B achieves only 42.5% average accuracy.

#English text is being prepared.#Multimodal#Papers/Research

AAnt Group inclusionAI: New GitHub Repository12:47First-partyModel releases

AI score59

Ant Group's inclusionAI Open-Sources 2B Visual Forensics Model DefakerOne2

Ant Group's GuangJian team has open-sourced DefakerOne2 on GitHub, a 2B-parameter unified visual forensics model built on the Qwen3.5-2B multimodal backbone, covering AI-generated content, DeepFake faces, document tampering, natural image tampering, face and ID anti-counterfeiting, and AI-generated video detection. The model was trained on approximately 50 million samples, employs a three-stage Tide-Style Tuning process from LoRA to full parameters and back to LoRA, and uses entropy-based test-time routing to select between the video and frame pathways. The team claims that it achieves SOTA on 40 of 53 detection benchmarks and on 11 of 14 localization benchmarks, and that the weights and the RRVBench benchmark have been released on ModelScope.

#Model releases#Open-source ecosystem#Multimodal

HHugging Face12:47First-partyModel releases

AI score60

NVIDIA Releases Nemotron 3.5 Transcribe and Granary v2 Data Pipeline

NVIDIA has released the open-weight model Nemotron 3.5 Transcribe on Hugging Face and ModelScope, integrating transcription, speaker diarization, translation, and audio question answering into a single checkpoint. It supports transcription in 40 languages, translation between 31 languages and English, and speaker labeling in 13 languages, can process recordings up to 1 hour long, and can label up to 8 speakers. The model ranks first among open-weight models on the English leaderboard of the Hugging Face Open ASR Leaderboard with an average WER of 5.33%, and achieves 2.8% AA-WER on Artificial Analysis with approximately 4B activated parameters. The simultaneously released Granary v2 covers 48 languages and 463,062 hours of speech data, and the accompanying NeMo Speech fine-tuning recipe reduced WER on held-out speakers from 22.44% to 12.17% in a medical case.

Same story · 9 reports

#Model releases#Speech#Open-source ecosystem

HHugging Face12:47First-party

AI score29

AnySpeechRecognizer and AnyDecisionModel: Open-source speech and decision models for the Apple platform

AnySpeechRecognizer supports engine switching on top of the Apple SpeechAnalyzer API, with options including NVIDIA Parakeet TDT 0.6B v3 (supporting 25 European languages), Moondream post-trained Parakeet Ultra, the English version Parakeet Unified, and Alibaba Qwen3-ASR (0.6B/1.7B, covering 30 languages and 22 Chinese dialects). These models have been converted to Core ML and published on the Hugging Face Hub; Parakeet takes about 20–40 seconds to transcribe 1 hour of audio, while Qwen3-ASR is slower and requires several GB of memory. The package also adds a speaker diarization module, SpeakerDiarizer, which the Apple framework lacks; it is based on NVIDIA Streaming Sortformer and can track up to 4 speakers.

#English text is being prepared.#Speech#English text is being prepared.

HHugging Face12:47First-party

AI score50

Hugging Face Announces Project ProtoStar: Making Publicly Funded Scientific Data AI-Ready

Hugging Face announced Project ProtoStar to promote the systematic sharing of publicly funded scientific research outputs in AI-ready formats, accompanied by open infrastructure and release guidelines. Hugging Face already hosts more than 17,000 public datasets in core scientific fields, including nearly 29,000 in medicine and mathematics; the number of new scientific datasets added in the first nine months of 2026 is more than triple that of the same period in 2025. The first challenge focuses on climate risk, using historical weather data from 210 U.S. cities through 2018 to predict actual climate conditions from 2019 to 2026.

#Industry news#Open-source ecosystem#English text is being prepared.

HHugging Face12:47First-party

AI score30

Granite Switch 4.1.1: Update Models Like You Update Software

IBM has released Granite Switch 4.1.1, advocating for updating models the same way you update software. The article covers three models: granite-switch-4.1-30b, granite-speech-5.0-470m-turboctc, and granitelib-core-r1.0, and features a Space demo, "Speech to Decisions with Granite," in which a single Granite Switch model listens to and evaluates speech in real time.

#Product updates#Open-source ecosystem#Hugging Face

TThe Beijing News12:47

AI score26

Fliggy Travel AI "Fliggy Bangbang" sees 118% surge in Double 11 interactions, launches cross-product price comparison

Fliggy Travel AI "Fliggy Bangbang" saw a 118% week-over-week increase in Double 11-related interactions over the past week, and launched a "Double 11 Assistant" that supports cross-product price comparison. The feature has been integrated with all Double 11 promotional products, and one-click room booking has begun gray-scale testing, covering hotel packages and hotel-plus-attraction packages. Users can describe their needs in natural language, and the AI matches products based on context such as membership tier and provides follow-up fulfillment services.

#Product updates#Agents

TThe Beijing News12:47

AI score29

Xinyan Robotics Releases Embodied Family Agent Bubbo 1

Xinyan Robotics officially released its embodied family agent "Bubbo 1" on October 29, delivering companion-style interaction based on its self-developed Bubbo OS. Founder Ren Yongliang said that technology determines the lower bound of the experience while interaction determines the upper bound, and expressed the hope of giving autonomy back to children and space back to families. The company was founded in 2025 and focuses on the research and development of emotional intelligence robots for the home.

Same story · 2 reports

#Product updates#Embodied AI#English text is being prepared.

TThe Beijing News12:47

AI score24

China Fortune Media Group and Beijing Property Rights Exchange Launch Vertical Intelligent Agent for State-owned Asset Property Rights Trading

On October 29, China Fortune Media Group and the Beijing Property Rights Exchange each launched a vertical intelligent agent for state-owned asset property rights trading. Among them, "Guozi Bao AI" was jointly developed by China Fortune Media Group together with Zhongchantou Technology and Zhipu, and features intelligent retrieval of state-owned asset property rights projects, multi-dimensional value mining, precise matching of investors, and intelligent interpretation of policies. The Beijing Property Rights Exchange's "Beijiao Huitou" platform has gathered more than 50,000 investment institutions and served more than 3,200 projects.

#Industry news#Agents

BBeijing News12:47Analysis

AI score67

Beijing News tests Qwen Go! and Lingguang 2.0: personal agent competition shifts toward autonomous task handling

Beijing News Shell Finance reporters tested Alibaba's Qwen Go! and Ant's Lingguang 2.0. The former set up a daily 8 a.m. medication reminder with a single sentence and checked it item by item, while the latter, after authorization on Alipay, sorted out the past six months of bills. The article also reviews the layouts of Meta Muse, OpenAI dots, as well as ByteDance and Tencent, and points out that cross-ecosystem permissions are currently the hardest wall to break down, with ByteDance's SAEP published rejection list reaching 155 apps.

#English text is being prepared.#Agents#OpenAI

TThe Beijing News12:47

AI score60

SK Hynix's Third-Quarter Revenue Exceeds 1 Trillion Won for the First Time, and It Announces a Shareholder Return Plan of About 1 Trillion Won

On October 28, SK Hynix released its financial report for the third quarter of fiscal year 2026, with revenue of 104.6155 trillion won, exceeding 1 trillion won for the first time, operating profit of 83.3338 trillion won, up 632% year on year, and an operating margin of 80%. On the same day, the company announced a shareholder return plan of about 1 trillion won for fiscal year 2026, including more than 30 trillion won in cash dividends and more than 60 trillion won in share buybacks and cancellation. The company plans to expand HBM4 production before the end of the year and said that long-term supply agreements have already covered more than 50% of total capacity for 2027 and 2028.

Same story · 2 reports

#Industry news#English text is being prepared.

TThe Beijing News12:47

AI score50

Aitu Technology Releases the Aini Robot and Production Line King, Entering the Field of Embodied Intelligent Sewing for Apparel

On October 28, Aitu Technology, a subsidiary of Jack Technology, released the embodied intelligent sewing robot Aini Robot and the apparel robot complete production line Production Line King. The first-release price of Aini in China is 308,000 yuan. Aini has achieved autonomous sewing in five core processes at Chenfeng Group: shirt collars, cuffs, patch pockets, buttonholes, and button attaching; Production Line King has entered the factory of Shengtai Group. Under the same production capacity, a traditional production line requires 40 people, while Production Line King requires only 12 people, and the project investment payback period is less than 3 years.

#Product updates#Embodied AI