AI Weekly Digest: the week's big stories (Week 46, 2026 · 11.09 — 11.15)
URAXAI
Google releases a high-performance agent model, AWS launches Claude model, Anthropic IPO oversubscribed, NVIDIA showcases robotic simulation technology, DeepMind releases AlphaGenome 2 model.
DeepMind Releases AlphaGenome 2 Model, Decoding How Cells Read DNA to Better Understand Diseases
DeepMind has released the AlphaGenome 2 model, which predicts how DNA sequences and their variations affect specific human cell types, providing scientists with a powerful new tool to reveal the cellular drivers of diseases.
Model releases & updates
In this sectionGoogle releases Gemini model series, DeepMind releases Gemini Robotics 2.5, AWS launches Claude model.
Google releases Gemini 4 Hydrogen agent, outperforming larger Argon model
Google has released the new agent model Gemini 4 Hydrogen, designed to handle daily tasks quickly and economically. The model outperforms the larger Argon model at the default 'high' thinking level and provides four thinking levels. It has a context window of one million tokens and both input and output are one million tokens.
NVIDIA showcases progress in robotic simulation technology at CoRL
NVIDIA showcased its latest progress in the field of robotic simulation at CoRL, including world action models, simulation tools, benchmarks, and agent workflows, aimed at advancing the development of robotic technology.
DeepMind releases Gemini Robotics 2.5, accelerating real-world general dexterity
DeepMind has released Gemini Robotics 2.5, an enhanced visual-language-action (VLA) model that can improve the breadth and speed of robotic learning skills, enabling it to assist humans in completing more tasks in more realistic environments.
Google releases Gemini 4 Carbon and introduces the 'swarm' feature for agents
Google released its latest AI model Gemini 4 Carbon and introduced the 'Flock' agent feature, allowing users to convert a single prompt into a coordinated team of intelligent agents. The Carbon model is competitively priced, starting at $2 per million input tokens and $10 per million output tokens.
BotPsy: Device-side Personal Assistant Model Based on Qwen3.5-0.8B Released
Tether AI Research released BotPsy, a device-side personal assistant model based on Qwen3.5-0.8B, suitable for tasks of low to medium difficulty, with excellent tool usage and agent performance.
Product releases & updates
AWS Launches Claude Fable 5.2 and Claude Mythos 5.2 Models
AWS launched Claude Fable 5.2 and Claude Mythos 5.2 models, targeting technical work and fields of cybersecurity and life sciences, respectively. Fable 5.2 has higher capabilities in coding and scientific fields, while Mythos 5.2 is verified through Anthropic's verification program. The models can be used on Amazon Bedrock and Claude Platform on AWS.
AWS Releases Next-Generation AgentCore Runtime, Enhancing Agent Security and Efficiency
AWS released the next-generation AgentCore runtime, designed to enhance the security and efficiency of agents. The new version provides faster startup times, elastic scaling, and a usage-based economic model, while ensuring isolation and data security.
AWS Announces New Features for AI Builders in October 2026
AWS announced new features and updates such as Amazon Bedrock AgentCore and Strands, aimed at helping developers build, connect, and optimize agents, and provided new models and pricing options.
Hugging Face Releases llama.cpp v1.0.0: A Stronger Foundation for the Local AI Ecosystem
Hugging Face released llama.cpp v1.0.0, a local AI platform that supports various hardware and model architectures, aiming to provide a lightweight, flexible, and stable API for building cross-platform and private AI products.
NVIDIA Nemotron 3.5 Super VL is now available in Amazon SageMaker JumpStart
NVIDIA Nemotron 3.5 Super VL, a visual language model, is now available in Amazon SageMaker JumpStart. The model supports image, video, and text input and generates text output.
NVIDIA releases OpenShell and new AI infrastructure
NVIDIA released OpenShell, an open-source runtime environment for secure and governable autonomous agents, which has been contributed to the Cloud Native Computing Foundation.
Industry news
In this sectionAnthropic IPO oversubscribed, NVIDIA CEO resigns, AWS launches reusable asset acceleration for generative AI delivery.
Anthropic IPO oversubscribed, becoming one of the largest public offerings in history
Anthropic's IPO was oversubscribed six times, with a target valuation of about 1.4 trillion dollars, becoming one of the largest public offerings in history.
AIE CODE event: multiple AI models released and progress
In the AIE CODE event, Google released the Gemini Decision 1 model, Cloudflare showcased the Clef and Clairic models, and introduced the latest progress and evaluation results of multiple AI models.
NVIDIA CEO Resignation and Anthropic IPO Spark Industry Attention
The resignation of the NVIDIA CEO and the ongoing IPO of Anthropic have sparked industry attention.
Chinese AI Models like DeepSeek May Use Banned Nvidia Chips, Raising US Concerns
A study finds that leading AI models developed by Chinese companies may have used Nvidia chips restricted for export by the US government, indicating challenges for Nvidia in chip design. Models released by companies like DeepSeek and Moonshot AI may be affected by the embargoed chips. Epoch AI estimates that about 2 million advanced chips will enter China in 2024 and 2025, including 275,000 Nvidia AI accelerators entering Malaysia.
AWS Launches Reusable Assets to Accelerate Generative AI Delivery
AWS has launched reusable assets aimed at accelerating the delivery of generative AI solutions. These assets include retrieval-augmented generation (RAG) pipelines, document processors, agent frameworks, and more, which can help developers save development time and improve work efficiency.
Research papers
In this sectionDeepMind Releases AlphaGenome 2 Model, Microsoft Researchers Discover AI Model Backdoor Technique 'Lazarus'
DeepMind Releases AlphaProof Code: Breakthrough in Deep Verification of Complex Software
DeepMind researchers have developed a new intelligent agent framework called AlphaProof Code for deep formal verification of complex software. This framework combines human expert knowledge with AI agents, and for the first time uses AI to perform deep formal verification on widely deployed cryptographic protocols, bringing a breakthrough in the field of software security.
Microsoft Researchers Discover AI Model Backdoor Technique 'Lazarus'
Microsoft researchers have discovered a backdoor technique named 'Lazarus' that can reprogram dormant neurons in AI models, causing normal AI agents to execute malicious commands when hidden triggers are present. This research reveals new issues in the field of AI security and proposes corresponding defensive measures.
OpenAI Research Reveals Hidden Thought Processes of Advanced Models
OpenAI's research indicates that advanced models like GPT-6 Astra are becoming difficult to monitor because developers' optimizations in performance and cost lead the models to hide their thought processes.