Astra for Law leads in legal intelligence
Astra for Law ranked first in a launch evaluation spanning 24 areas of law. Every ChatGPT Enterprise customer can now request access.
Astra for Law ranked first in a launch evaluation spanning 24 areas of law. Every ChatGPT Enterprise customer can now request access.
OpenAI wants GPT-6 Astra and GPT-6 Luna to automate administrative tasks in US healthcare through computer-use agents that operate existing clinic software and insurance portals the way a human would. Luna handles tasks like prior authorizations for about 8.5 cents each. Companies including CVS Health and Cigna Healthcare are already deploying the technology. The article AI agents are starting to fill out medical forms and file insurance requests in place of human staff appeared first on The Decoder .
NVIDIA’s Ian Buck kicked off the company’s week at SC26 in Chicago on Monday with new tools for AI agents, software that can plan experiments, run code and check results on researchers’ own systems.
Why it matters: The article highlights NVIDIA's advancements in AI agents, high-performance computing, and quantum computing, showcasing the company's commitment to pushing the boundaries of scientific research.
Cloudflare made its name speeding up the web. Now it's building System One models to make decisions and pull data in milliseconds.
OpenAI extends its ChatGPT for Academic Researchers program, granting 200,000 researchers global free access to its advanced models. This initiative aims to broaden AI accessibility for scientific research.
ChatGPT will ban an account trying to access CSAM content for 24 hours as a warning.
Why it matters: OpenAI's new tools for combating CSAM highlight the company's commitment to addressing safety concerns in AI, offering insights into the potential impact of AI technology on society.
Alibaba releases the new Qwen4-Max intelligent agent, adopting a more efficient architecture and lower prices for professional work in industrial, scientific, and financial fields. The hybrid expert model has 240 billion parameters, but each token only activates 71 billion parameters.
Why it matters: The original text provides changes in capabilities and open entry, allowing readers to judge how it will change the existing workflow.
AI coding agents can waste time and tokens relearning the same lessons. TokenEconomizer is an open-source plugin that builds on past sessions to cut costs while maintaining task success. It’s a path toward coding agents that learn not only how to solve tasks, but also how to spend computation more effectively.
Why it matters: TokenEconomizer is an innovative open-source plugin that significantly reduces the token consumption of AI coding agents by learning from past sessions, offering a practical solution for developers and researchers in the AI coding space.
The Amazon Bedrock Managed Knowledge Base now supports table search, allowing users to retrieve answers from tables through intelligent retrieval. When questions require data from CSV, XLSX, or XLS files, table search will look for values in one or more tables or worksheets and perform calculations.
The Novo Nordisk manufacturing technology team utilizes agentic AI on the AWS platform to develop an intelligent support system aimed at reducing document update time, improving document visibility, and making key manufacturing knowledge more accessible. The system is based on the GraphRAG architecture, combines knowledge graphs and mixed search, and is built on the Amazon Bedrock AgentCore platform, utilizing the Anthropic Claude model to provide accurate answers.
Hugging Face and OpenRouter jointly launch OpenRouter to simplify the operation of open-weighted models. All major open-weighted models have been published on the Hugging Face Hub, with OpenRouter providing services at scale, routing each request to over 80 providers based on real-time measurements. Now, language model requests will go through OpenRouter, selecting the best provider based on real-time price, speed, and reliability, and retrying on another provider if the first one fails.
Why it matters: The original text provides changes in capabilities and open entry, allowing readers to judge how it will change the existing workflow.
NVIDIA celebrates the 20th anniversary of the CUDA platform, looking back at the contributions of GPU computing in the fields of science and AI, including promoting Nobel Prize-winning AI research and accelerating scientific applications on supercomputers.
Why it matters: The development history of the CUDA platform showcases the significant impact of GPU computing in the fields of science and AI, allowing readers to understand the development trends and future potential of GPU computing.
NVIDIA has released the expanded CUDA-X libraries, including cuPhoton, cuPotential, and cuSpa, which can help researchers accelerate their research from astrophysics and materials science to single-cell biology. These new libraries provide tools for AI agents to help researchers make discoveries faster than ever before.
Why it matters: The original text provided changes in capabilities and open entry, allowing readers to judge how it will change the existing workflow.
NVIDIA has released a new Slurm version that supports quantum processors, making it easier for supercomputing centers to integrate quantum computing into their workflows.
Why it matters: The original text provides changes in capabilities and open entry points, allowing readers to judge how it will change the existing workflows.
Lambda is using NVIDIA Photonics' CPO switches to connect NVIDIA GB300 NVL72 racks, achieving connections for over 10,000 GPUs, providing lower power consumption, higher resilience, and longer average interruption intervals. NVIDIA Photonics provides better power efficiency and reliability than traditional pluggable optical modules by using co-packaged optical engines on switch packages.
Why it matters: The original text provided changes in capabilities and open entry, allowing readers to judge how it will change the existing workflow.
Audible has launched an AI-assisted summary feature that helps users determine if they will like a book before they start listening, using audiobook samples, AI summaries, and author comments. The feature aims to prevent users from researching titles on other platforms and has been launched in some English audiobook titles with plans to expand to other regions. Additionally, Audible is using AI to personalize audiobook recommendations for users and has expanded two test features: Ask a Question and Recaps.
Dreamteam, a startup founded by former Freshworks executives, has raised $53 million in funding to launch an AI native CRM aimed at challenging Salesforce and HubSpot. The company aims to eliminate much of the work in traditional CRM by automatically capturing contextual information from sources such as emails and customer conversations, as well as using AI agents to handle tasks such as follow-ups, meeting preparation, and pipeline monitoring.
Apple announced a collaboration with OpenEvidence to conduct research aimed at improving the effectiveness of AI health tools. The research collects user feedback on medical issues through an iOS app and evaluates the accuracy, detail, and relevance of AI tools. The app is now available for download, recruiting about 200,000 participants, with the aim of improving the design and functionality of AI health tools to make them easier for users to understand and take action.
Anthropic's Claude Fable 5.2 simulates tens of thousands of brushstrokes on a virtual canvas using code to create complex and even realistic paintings. The model does not require external image models or pre-rendered images, controls the brushstrokes with Python code, and can self-correct like a real painter.
Meituan releases the new agent model LongCat-2.5, with a parameter count of 1.6 trillion, fully trained on Chinese hardware, and achieves a level comparable to Western models such as Claude Opus 5 in coding benchmark tests. The model supports processing text, images, and videos, and allows users to turn on or off the inference function.
Why it matters: The original text provides changes in capabilities and open entry points, allowing readers to judge how it will change the existing workflow.
OpenAI has redistributed its robotics research team members to other teams because general artificial intelligence models like GPT-6 Astra may be able to directly control robots, reducing the need for specialized robot AI.
Runway has released its robot model Praxis-1 and provided open-source weights and source code. The model uses knowledge learned from videos to control robots and has shown promising results in the company's own tests. Praxis-1 controls robots through video training and reduces the need for expensive robot-specific datasets traditionally used to train such systems.
NVIDIA NeMo Switchyard is an open-source library that helps AI agents choose the model to process each request. This article introduces how to deploy LiteLLM, an open-source large language model (LLM) gateway, using AWS Cloud Development Kit (AWS CDK) and use the Switchyard routing plugin to optimize the routing of Claude Code and OpenAI Codex CLI requests.
Amazon SageMaker JumpStart has released OpenFold3, an open-source biomolecular structure prediction model. Users can deploy it to their own AWS account with just a few lines of code and predict the 3D structures of proteins, DNA, RNA, small molecule ligands, and their complexes.
Why it matters: The release of OpenFold3 on Amazon SageMaker JumpStart provides new tools for biomolecular structure prediction, allowing readers to understand how it changes existing workflows.
Google Research introduces the/Teamwork feature, utilizing the Gemini 4 model and multiple agent workflows to achieve new progress in information theory, formal verification, molecular structure discovery, and kernel optimization, and is open to researchers and developers.
Why it matters: The original text demonstrates the application of/Teamwork in multiple fields, including mathematics, science, complex systems engineering, and molecular chemistry discovery, allowing readers to understand its potential applications in these areas.
OpenAI launches ChatGPT Wallet, allowing users to shop through Dots agents. The service is currently available only to US ChatGPT Pro subscribers and requires binding a credit card that meets the criteria. When shopping, the agent will confirm with the user first and ensure security through biometric verification. OpenAI plans to allow users to fully delegate shopping to agents in the future and set clear parameter limits.
AssemblyAI has achieved cost-optimized voice recognition inference on AWS using Amazon EC2 Spot instances. By mixing Spot instances with On-Demand instances and services like Amazon ECS and Amazon SQS, it has implemented a large-scale, low-cost, and reliable voice recognition service.
OpenAI launches ChatGPT wallet, allowing AI agents to make payments on behalf of users. This feature is currently available in the US through Dots and requires a ChatGPT Pro subscription and eligible US credit cards. OpenAI plans to eventually delegate all purchases to their agents, within set spending limits, product categories, and time frames.
AWS SageMaker HyperPod now provides data capture functionality, which can record real-time inference inputs and outputs and store them in Amazon S3. This feature helps detect model drift, identify failure patterns, and improve models using real interactions.
Why it matters: The original text provided changes in capabilities and open entry points, allowing readers to judge how it will change the existing workflow.
Anthropic has released Claude for Manufacturing, a plugin that integrates AI directly into industrial software to automate factory workflows. The solution connects to Autodesk Fusion, Oracle NetSuite, Microsoft 365, and Infor through four general interfaces.
OpenAI launches the Daybreak Security Product, including Daybreak Red Team and Daybreak Defense Factory, which are used for automatically detecting application vulnerabilities and automating defense.
Former OpenAI researcher Will DePue discovered that Chinese AI companies DeepSeek, Moonshot AI, MiniMax, and Ant Group are using NVIDIA's Blackwell GPU, which is subject to export restrictions. DePue's analysis provides the first technical evidence that Chinese AI laboratories can access NVIDIA's latest hardware despite U.S. export controls.
Hugging Face has released llama.cpp v1.0.0, a local AI platform that supports various hardware and model architectures, aiming to provide lightweight, flexible, and stable APIs for building cross-platform and private AI products.
Why it matters: The release of llama.cpp provides a stronger foundation for the local AI ecosystem, allowing readers to understand how it supports various hardware and model architectures, as well as its contributions to the open-source community.
ArtCraft company released PhotoCraft, an open source Photoshop alternative based on AI, aiming to rebuild the internet infrastructure and plans to develop more open source applications.
Natera has collaborated with AWS to develop SWARM, an intelligent agent system for laboratory operations using Amazon Bedrock AgentCore. The system supports ticketing through automation and rapid response, improving diagnostic accuracy and efficiency.
Why it matters: The original text provides changes in capabilities and open entry points, allowing readers to judge how it will change the existing workflow.
Anthropic launches the Claude Meetings feature, allowing AI to join video conferences as visible participants, answer questions, generate documents and whiteboards, and record notes and action items. The feature is currently in testing phase and supports Google Meet platform, suitable for Team and Enterprise plans.
Amazon announced the general availability of SGLang-Neuron, accessible via the AWS Deep Learning Container. SGLang-Neuron allows users to use the SGLang interface to provide models and generate reinforcement learning (RL) replays on AWS Trainium.
Google has launched the new Gemini 4 Hydrogen model, which is better at understanding and answering complex questions, including multi-part questions, difficult-to-understand typos, or highly specific requirements. The model has already started to be used in search, Gemini applications, all levels of Google AI Studio, Gemini API, Gemini for business, and Google Antigravity.
Nectar launches a new AI intelligent body aimed at automating the discovery, contact, collaboration management, and effect tracking of brand creators to alleviate the burden on marketing teams. The intelligent body can search over 300 million creator profiles, analyze content matching with brands, send personalized messages, coordinate collaborations, and monitor the effectiveness of activities.
Google has released the new Gemini 4 Hydrogen model, replacing the previous faster and more cost-effective Flash model. The model has significant improvements in speed and accuracy, running 2.5 times faster than Flash 3.8 and even comparable to the Argon model in some tests.