Anthropic launched Claude Cowork on web and mobile Tuesday for Max subscribers, allowing tasks to run in the background across devices. Usage data from 1.2 million sessions shows business operations account for 33.4% of Cowork tasks, while software development represents just 8.7%.
Anthropic is expanding Claude Cowork from desktop-only to mobile and web platforms. The autonomous agent feature will run tasks in the background across devices, with scheduled execution requiring no device to be online, starting with Max plan subscribers in a gradual rollout over the next several weeks.
Microsoft announced at Build 2026 that Foundry Managed Compute now includes a curated catalog of open-weight models from Hugging Face's 3 million model repository, refreshed weekly with one-click deployment. The service pre-stages weights in Azure, provides Microsoft-scanned runtimes (vLLM, SGLang, TensorRT-LLM, NIM, TEI, llama.cpp), and offers pay-per-hour GPU pricing with automatic security patching.
Chinese AI models including DeepSeek and Z.ai have captured over 30% of weekly token usage by U.S. companies on OpenRouter since February 2025, up from 4.5% in the first half of the year. The shift comes as companies seek alternatives 60-90% cheaper than leading models from OpenAI and Anthropic, while Chinese models close the performance gap to within 6-9 months of U.S. frontier systems.
AWS has developed Reverse Direct Preference Optimization (rDPO), a novel unlearning technique that reduces over-deflection in Amazon Nova models by up to 53 percentage points. The approach allows organizations to selectively adjust content moderation safeguards while preserving general model capabilities through LoRA adapters.
Nex AGI has released Nex-N2-Mini, an open-source agentic mixture-of-experts model with a 262K-token context window. The model accepts text and image inputs and is priced at $0.025 per 1M input tokens and $0.10 per 1M output tokens.
Amazon Web Services has added three MiniMax models to Amazon Bedrock: M2, M2.1, and M2.5. The newest model, M2.5, uses a mixture-of-experts architecture with 230 billion total parameters and 10 billion active per token, trained specifically for agent-native execution and coding tasks.
AWS has released infrastructure for deploying multi-turn reinforcement learning to train Amazon Nova models on SageMaker HyperPod. The system requires a minimum of 10 ml.p5.48xlarge instances and costs approximately $786-$1,180 per hour when running.
AWS has released an automated pipeline for redacting personally identifiable information in images, using Amazon Nova 2 Lite as an intelligent coordinator. The solution combines Nova's contextual vision reasoning with Meta's SAM 3 model deployed on SageMaker and Amazon Textract to handle complex PII detection scenarios including faces, fingerprints, ID cards, and license plates.
Anthropic has released an official Linux desktop app for Claude, bringing feature parity with macOS and Windows versions. However, the app currently fails to reliably connect to locally-installed AI models like Ollama, limiting Linux users to cloud-based Anthropic plans.
Tencent has released Hy3, a 295-billion parameter Mixture-of-Experts model with 21 billion active parameters and a 256,000-token context window. The model features configurable reasoning modes and is available free through OpenRouter, with deployment ending July 21, 2026.
Tencent has released Hy3, a 295-billion parameter Mixture-of-Experts model with 21 billion active parameters and 3.8 billion MTP layer parameters. The model features a 256K context window and is released under Apache 2.0 license, with pricing not yet disclosed.
Google has detailed pricing and features for its Gemini app subscription tiers. AI Plus costs $4.99/month and includes 128,000 token context windows, while AI Pro at $19.99/month provides 1 million token context windows. Free users are limited to 32,000 tokens.
Mistral AI has released Leanstral 1.5, an open-source 119B parameter mixture-of-experts model designed specifically for Lean 4 proof assistance. The model features 128 experts with 4 active per token (6.5B activated parameters), a 256k token context window, and multimodal input capabilities.
NVIDIA released Nemotron-Labs-TwoTower-30B-A3B-Base-BF16, a block-wise diffusion language model that generates text by denoising blocks of tokens in parallel rather than sequentially. According to NVIDIA, the model achieves 2.42× the wall-clock generation throughput of its autoregressive baseline while retaining 98.7% of aggregate benchmark quality.
Mistral AI released Leanstral 1.5, a free Apache-2.0 licensed model with 119B total parameters and 6B active parameters specialized for formal verification in Lean 4. The model achieves 100% on miniF2F benchmark, solves 587 of 672 PutnamBench problems at $4 per problem (versus $300+ for competitors), and reaches state-of-the-art 87% on FATE-H and 34% on FATE-X benchmarks.
Anthropic has launched the public beta of Claude Science, an AI workbench for scientific research that integrates NVIDIA's BioNeMo Agent Toolkit. The platform allows scientists to execute end-to-end research workflows using natural language commands to interact with digital agents.
Apple has released an MCP server for Safari Technology Preview 247 that allows AI coding agents to directly inspect and debug websites. The server gives agents access to console logs, network requests, screenshots, and DOM interactions through the Model Context Protocol standard created by Anthropic.
GitHub has added the Microsoft C++ Language Server as a plugin to the Copilot CLI marketplace. The plugin includes a built-in setup skill designed to automate C++ project configuration.
Amazon Bedrock now supports NVIDIA Nemotron and OpenAI GPT OSS models in AWS GovCloud (US) Regions. The launch includes OpenAI's GPT OSS models (120B and 20B parameters, 128K context) and NVIDIA Nemotron 3 family (9B to 120B parameters, 1M context), providing government agencies FedRAMP High and DoD SRG Level 5-compliant AI inference on U.S. soil.