All news

IEnglish text is being prepared.16:03Product release

AI score55

DeepSeek official website code reveals Cyber, an independent cybersecurity model, in grayscale testing split into enterprise and personal editions

Developers discovered the application entry point and API logic for the Cyber cybersecurity model in the front-end code of the DeepSeek official website, with model identifiers cyber-personal-v3 and cyber-enterprise-v4. The personal edition requires ID card information and completion of facial verification, while the enterprise edition additionally requires submission of a business license and authorization letter; both channels require a real-name account, registration in mainland China with a bound mainland mobile phone number, and no violation records within the past 90 days. Currently, this entry point does not appear in the public menu, and only a small number of grayscale accounts can see the relevant section on the usage page. DeepSeek has not yet issued an official announcement.

#Product updates#Safety / alignment#DeepSeek

PPhoenix Technology16:03Product release

AI score60

XHEART New Core Voyage Releases 700 TOPS Automotive-Grade Chip X7 Ultra

XHEART New Core Voyage has released the automotive-grade chip X7 Ultra, with 700 TOPS of computing power, focusing on reverse-defining chip architecture and toolchains based on large-model requirements. The previous-generation X7 series has cumulatively shipped more than 200,000 units, is deployed in 10 mass-production vehicle models, and has secured orders for more than 40 vehicle models. The company also announced that its next-generation X9 targets 3,000 TOPS of computing power and 700GB/s single-chip memory bandwidth, with launch planned for 2027.

Same story · 7 reports

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

336Kr · AI15:52Product release

AI score71

Neuralink Connects Grok to the Brain: ALS Patient Controls AI Agent with Thoughts

Neuralink has announced its latest progress: Kenneth, a patient with ALS, silently thinks "Hey Neuralink" and formulates a task in his mind, which then wakes up the Grok agent on his computer to autonomously execute the operation. The team conducted a 72-hour development sprint at the subject's home, producing 8 functional prototypes, including neural voice activity detection (VAD) that requires no manual switch, a system-level brain-computer input method called BCI Keyboard Dictation, and a mobile app called Bloom. Kenneth also used speech generated by the brain-computer interface to connect to the iPhone call audio stream, making a phone call to his wife with his thoughts for the first time in seven years.

Same story · 3 reports

#Product updates#Agents#Embodied AI

JJiwei.com15:52Product release

AI score65

Iluvatar CoreX's Tiangai 300 Benchmarked Against H200 in Real Tests, and On-Device Terminal Tongyang TY1200 Released

Iluvatar CoreX held its Shanghai media open day on October 28, where it opened the Tiangai 300 for live demonstrations for the first time, saying that in third-party real tests its single-card end-to-end throughput reached 1.05 times that of the H200. The company also provided lab data such as 64K-context Attention at 1.05 times that of the H200 and MoE throughput under GLM-5.3 at 1.15 times, and showcased its cloud-edge-device layout, with the on-device terminal Tongyang TY1200 rated at 300 TOPS INT8 dense computing power.

Same story · 4 reports

#Industry news#English text is being prepared.#Reasoning

Ggihyo.jp15:51Product release

AI score52

Gemini Notebook Can Add Select U.S. and India Textbooks for Free

On October 29, 2026, Google announced that Gemini Notebook (formerly NotebookLM) can add hundreds of textbooks for free, with sources including U.S. and Indian publishers such as OpenStax, Open Textbook Library, and S Chand Publishing. Google's Josh Woodward said on X that more than 700 textbooks are available for free. Users open the catalog by selecting Textbooks under Add sources within a notebook, can filter by subject, level, and country, and can ask questions based on textbook content, view citation links, and generate study guides, practice quizzes, and flashcards.

Same story · 5 reports

#Product updates#Google

SSina Tech15:02Product release

AI score52

Baidu upgrades search AI API, minimum 4.9 yuan per thousand calls, lower than DeepSeek

At a product upgrade communication meeting, Baidu App announced an upgrade to its search AI API, with a minimum price of 4.9 yuan per thousand calls during the promotional period, lower than the DeepSeek search API. The upgraded Wenxin Assistant moves search from a single question-and-answer exchange toward delivering finished outputs, covering more than 170 scenarios; AI Lens supports simultaneous recognition of multiple targets and dynamic tracking, and enables real-time conversation in 9 dialects. GUI interactive scenarios on the search results page have increased 5-fold, with access to over 5 billion data resources from government affairs, livelihood, the central bank and others, while AI interactive dramas and interactive applications have launched simultaneously.

Same story · 3 reports

#Product updates#Search#DeepSeek

WWallstreetcn15:02Product release

AI score58

Doubao Work Opens Third-Party Models Such as Kimi, GLM, and DeepSeek to Enterprise Customers

On October 29, ByteDance's Doubao Work announced that it would open third-party large model services to customers on its Team Subscription and Enterprise Subscription plans. The first batch of integrated models includes Kimi 3.1 Pro, GLM-5.3, and DeepSeek V4.1 Flash, among others. Enterprises can call them directly using their existing AI credits without needing to separately purchase API access. The article notes that Tencent WorkBuddy has already integrated models including Hunyuan, GLM, MiniMax, Kimi, and DeepSeek, while Qwen Office's model selector still primarily offers its own options such as Advanced, Basic, Economy, and Qwen3.8-Max. After integrating multiple models, platforms also need to coordinate different models' context lengths, tool calling, and costs at the Harness level; model aggregation and automatic routing are not the same thing.

Same story · 3 reports

#Industry news#Agents#English text is being prepared.

QQbitAI15:01Product release

AI score64

Genspark Releases Video Agent Workspace Media Studio, Integrating 70+ Models

Genspark has officially released its video Agent workspace Media Studio, with a large language model serving as director and integration of 70+ model versions from more than 20 vendors, covering video, image, audio, and 3D. The author tested using Seedance 2.5 to generate a video of Ultraman and Dario rapping on the same stage, and used the story bible feature to complete the business-war short drama "Pricing Night"; running 3 cases consumed about 45,000 credits. According to the official description, knowledge-explainer videos follow a code-to-video route, and code rendering itself does not consume credits, while narrative films offer 3D white-model previews and partial reshoots to reduce rework.

Same story · 2 reports

#Product updates#Agents#Video

QQbitAI15:01Product release

AI score63

1.3B Open-Source Document Parsing Model OXR Tops Three Leaderboards, Introducing RSI Self-Evolution Training

The domestic open-source document parsing model OXR, with 1.3B parameters, has topped three leaderboards: OmniDocBench, olmOCR-Bench, and DocParsingBench, and has already gained 1.1K+ Stars on GitHub. In the post-training stage, the model introduces a Recursive Self-Improvement (RSI) training paradigm for document parsing, using a general-purpose multimodal model to classify the parsing results of unlabeled documents into three categories—Correct, Wrong, and Uncertain—and then performing attribution clustering to generate targeted synthetic data that is fed back into training. Architecturally, it delegates layout analysis to an independent, pluggable lightweight model, while the main model is responsible for both content recognition and reading-order determination. A single RTX PRO 6000 takes 101.0 seconds to parse 100 pages end-to-end, and the code, weights, and evaluation reproduction pipeline have all been made public.

#Model releases#Open-source ecosystem#Multimodal

RRohan Paul14:31Product release

AI score54

Cartesia Releases Managed Agents, Ranking First in Blind Human Preference Tests on Speech Agent Arena

Cartesia has launched Managed Agents, an end-to-end voice agent platform for enterprises, which ranks first in blind human preference tests on Artificial Analysis's Speech Agent Arena and first on ServiceNow EVA-Bench. The platform uses a cascaded architecture, jointly tuning its proprietary voice models and the agent harness as a single system. Its H-Nets learn directly from raw bytes, achieving 3–5x efficiency gains in controlled scaling experiments, and at larger scale matches an open-source model trained on 11 trillion tokens with roughly 10x less training data and compute.

Quoted postCartesia · @cartesia

Introducing Cartesia Managed Agents: our end-to-end voice agent platform for enterprise. #1 on @ArtificialAnlys Speech Agent Arena and @ServiceNow EVA-Bench, built by co-designing our voice models and harness as one system. Check it out → https://t.co/7kJhCYqjQo https://t.co/NVOAds9czf

Same story · 2 reports

#Product updates#Speech#Agents

PPhoenix Technology14:01Product release

AI score62

XHEART Releases Large-Model Chip X7 Ultra with 700 TOPS of Compute

XHEART has released the X7 Ultra, a new-generation chip designed for billion-parameter-level smart-driving large models, with 700 TOPS of compute. The company says that in real-world tests using a partner's production Transformer smart-driving model, compute utilization reached 98.7%, and that while matching the model performance of industry flagship chips, memory bandwidth consumption dropped by 80% and inference energy efficiency improved by 100%. The X7 series has shipped a cumulative 200,000 units in 8 months of mass production, equipping 10 production vehicle models, with the first launch model being the SAIC Volkswagen ID.ERA 9X; at the same event, Momenta released the R7 Ultra world model, whose parameter count is 5 to 10 times higher than the previous generation.

Same story · 7 reports

#Product updates#English text is being prepared.#Reasoning

PPhoenix Technology14:01Product release

AI score69

Neuralink Connects Grok to the Brain: ALS Patient Controls AI and Makes a Phone Call With His Thoughts

Neuralink has announced its latest progress: Kenneth, a patient with ALS, silently thinks "Hey Neuralink" and states a task in his mind, which wakes up the Grok Agent on his computer to autonomously carry out operations. The team conducted a 72-hour development sprint at the participant's home, producing 8 technical prototypes, including neural voice activity detection that requires no manual switch, a system-level brain-computer input method called BCI Keyboard Dictation, and a mobile app called Bloom. Kenneth also connected speech decoded by the brain-computer interface to the iPhone call audio stream, and after seven years, made a phone call to his wife using his thoughts.

Same story · 3 reports

#Product updates#Agents#Embodied AI

CcnBeta14:00Product release

AI score71

Neuralink Subject Uses Thoughts to Wake Grok Agent and Control Computer

During a 72-hour hackathon, the Neuralink team developed 8 functional prototypes for P21 subject Kenneth, including the ability to silently think "Hey Neuralink" in his mind and think through a task to trigger the Grok agent to operate the computer autonomously. The team also rewrote the neural voice activity detection system to achieve pure thought-based speech decoding without button presses, and routed the decoded speech into the iPhone call microphone audio stream, enabling this ALS patient to call his wife using only his thoughts. In addition, the dedicated app Bloom supports placing orders by thought in a café, and a system-level brain-computer input method allows text entry in any app.

Same story · 3 reports

#Product updates#Agents#Speech

RRohan Paul12:53Product release

AI score53

Claude for Legal Adds a Library of Prebuilt Mini-Apps Covering Citation Checks and Contract Clause Review

Claude for Legal has added a batch of prebuilt mini-apps for everyday legal work such as verifying citations and sources and flagging contract clauses that violate company rules, some of which were built by Harvey, PwC, Goodwin, and Foley & Lardner. After opening an app, lawyers can first see how it performs on sample data, then point it at their own files and let Claude complete the first round of processing, with every finding linked to its source. Claude's official account said the app library also covers scenarios such as contract review and regulatory horizon scanning; users can start from the built-in examples, add their own documents and connectors, and share them across the enterprise.

Quoted postClaude · @claudeai

New in Claude for Legal: a gallery of ready-made artifacts for contract review, regulatory horizon scans, cite checks, and more. Start from the built-in example, add your own documents and connectors, and share it with anyone across the business. https://t.co/Wn2VK2HeMU

Same story · 3 reports

#Product updates#Agents#English text is being prepared.

GGuillermo Rauch12:53Product release

AI score60

Vercel CEO Explains Vercel CI in Detail: Infrastructure, Turborepo Integration, and Agentic CI

Vercel CEO Guillermo Rauch published a post explaining how Vercel CI solves the AI coding bottleneck, in three layers: fast and stable infrastructure (20 clusters worldwide, running millions of microVMs per day), Turborepo integration (automatically computing the dependency graph and skipping and caching existing work), and Agentic CI. He said that vercel ci in the CLI is the first step; agents can now inspect, abort, trigger, and replay CI, and Vercel Agent can assist with optimization. More capabilities will be added later, giving CI engineer-like judgment. The official Vercel account previously announced that Vercel CI had entered beta, with CI in internal repositories sped up 2-6×, and optimized for agents via Vercel MCP and the CLI.

Quoted postVercel · @vercel

Introducing Vercel CI • Intelligently skips tasks with Turborepo • 2-6× faster CI in our internal repositories • Optimized for agents via Vercel MCP and CLI After nearly 8 million tasks, we're excited for the future of agentic CI. In beta on all plans. https://t.co/IcsuvweAhB

#English text is being prepared.#Agents#English text is being prepared.

RRohan Paul12:53Product release

AI score50

Sunra launches a unified API to call image, video, audio and 3D models at the vendors' original prices

Sunra has released Sunra API, which lets users call all major image, video, audio and 3D models with a single key, billed at each model vendor's own API price, on a pay-as-you-go basis, with no subscription fee and no markup. Rohan Paul relays that Seedance 2.5 costs half as much on Sunra as on two other gateways, and that this unified entry point also makes it easy to run A/B tests on models from Google, OpenAI, ByteDance and Kling.

Quoted postSunra · @sunra_ai

Today we're launching the Sunra API. Every leading image, video, audio and 3D model through one key, each at its maker's own API price. Pay as you go, with no subscription and no markup. The models are ready. Your app is next. https://t.co/bQzsZ8DJZy

#Industry news#Video#English text is being prepared.

SShao Meng12:53Product release

AI score65

The MiniMax Code team open-sources MCode Stack Agent workflows and Skills

The MiniMax Code team has open-sourced MCode Stack, writing the Agent workflows and Skills the team uses daily into installable SKILL.md files and manifests, which can be integrated as plugins into MiniMax Code and any Agents that support Skills. The repository packages stacks by member, among which stacks/ronny/ is the most complete, containing 15 skills covering sub-Agent decomposition, context and tool budget auditing, long-session handoff, independent review, and hotfix judgment. The companion notes/ column MCode Engineering Notes explains the rationale; the six engineering-* entries are adapted from Google Engineering Practices, with the adapted portions under CC-BY 3.0 and the rest under MIT.

Quoted postRonnyHe · @Ronny_MiniMax

We just open-sourced MCode Stack: the actual workflows and skills the MiniMax Code team uses every day to build MiniMax Code. https://t.co/vREt7i8m2M

Same story · 2 reports

#English text is being prepared.#Agents#Open-source ecosystem

TTesting Catalog12:53Product release

AI score64

Anthropic's Claude Duet opens to US Max users, with Pro users to follow

Anthropic's Claude Duet is opening to US Claude Max users, and US Pro users will receive invitations afterward. Expansion outside the US is planned but has no timeline yet, and EU users may wait longer. Karan Ahuja said the team is prioritizing getting US Max users off the waitlist, and more invitations will go out this week.

Quoted postkaran ahuja · @ahuja_karan

Day 6 of Claude Duet! A few updates on access, and really appreciate everyone’s patience on the waitlist: - We’re getting all Max users in the US off the waitlist first. Lots of invites went out yesterday and today, with more coming this week. - Right after Max, we’ll start inviting Pro users in the US. - For folks outside the US, we’re working hard on extending access and will share a date as soon as we have one. If you’d like to try Claude Duet and help shape it, please join the waitlist. https://t.co/ucxIrHrNaa We’re building it with and for working professionals, small businesses, and entrepreneurs who are constantly juggling a lot. Team is reading every piece of feedback and shipping daily. More soon on how we built Claude Duet, how people are using it, new features and improvements.

#Product updates#Anthropic

TTesting Catalog12:53Product release

AI score75

Axiom Open-Source Repository: Dozens of AI Agents Generate About 43 Million Lines of Lean Code in 24 Days

Axiom has released an open-source GitHub repository in which dozens of AI agents converted the 15,000-page classification of finite simple groups proof into about 43 million lines of code within 24 days. The code is written in Lean and allows the proof to be verified step by step; the entire result is validated by running the code rather than by rereading hundreds of papers. Axiom's chart tracks the scale of AI-written Lean projects since mid-2025, and by its estimate the scale doubles roughly every 37 days.

Quoted postLeo · @biosemiote

The formal proof of CFSG may now be among the largest open-source projects ever. Watching agents assemble it is a reminder of the awe-inspiring human achievement behind it. The graph represents some 15,000 pages of intricate arguments developed by mathematicians over decades. https://t.co/13THDKI50f

Same story · 6 reports

#English text is being prepared.#Agents#English text is being prepared.

RRohan Paul12:53Product release

AI score77

Microsoft Releases GitHub Copilot Fusion, Routing Coding Tasks Between Claude and GPT Models

Microsoft has released GitHub Copilot Fusion, which can route coding tasks between Claude and GPT models. On Artificial Analysis's Coding Agent Index, its Max setting scores 60, versus 62 for Claude Code; the per-task cost is $5.59 versus $11.42. Fusion consumes 7.2M tokens per task, about half of Claude Code's usage.

Quoted postSatya Nadella · @satyanadella

Just wrapped our earnings call. It was a fast start to our fiscal year. Microsoft Cloud revenue surpassed $63B, with growth accelerating to 30% YoY. Lots of innovation as we build the new tech stack for the agentic wave. A few highlights:

Same story · 6 reports

#Product updates#English text is being prepared.#Agents