Anthropic just rolled out four beta tools on its API for autonomous agents: in-sandbox Python code execution, native remote MCP connections, persistent Files API storage, and prompt caching extended to an hour. It basically lets Claude crunch data and call third-party tools across long tasks without custom infrastructure.
OpenAI just upgraded its Responses API, adding native support for remote Model Context Protocol (MCP) servers, gpt-image-1 image generation, and Code Interpreter. Reasoning models like o3 and o4-mini can now call tools directly inside their chain-of-thought while preserving context tokens. OpenAI also officially joined the MCP steering committee to back the open standard.
Mistral just unveiled Devstral, a 24B coding model built with All Hands AI under a fully commercial Apache 2.0 license. It fits on a single RTX 4090 or a 32GB Mac, beating Google's Gemma 3 27B and DeepSeek's V3 on SWE-Bench Verified. Why it matters: Developers finally get a capable, agent-ready code model they can run locally without restrictive licenses.
NVIDIA just released Llama Nemotron Nano 4B, an open reasoning model tailored for edge agents. It delivers higher accuracy and 50% faster throughput than other sub-8B open models across coding, math, and function calling. It runs directly on Jetson and RTX GPUs for low-cost local reasoning.
Google just introduced Google AI Ultra, a premium $249.99/month subscription offering its highest usage limits and top-tier models. Subscribers get early access to Veo 3, the new Flow filmmaking tool, Project Mariner agent, and 30 TB of storage, while existing AI Premium plans are rebranded to Google AI Pro.
Google just launched Jules into public beta with no waitlist. Powered by Gemini 2.5 Pro, the autonomous agent clones your GitHub repos into secure cloud VMs to fix bugs, write tests, and submit PRs asynchronously in the background. It even reads aloud audio changelogs, turning routine backlog chores into hands-off tasks.
Google just introduced an AI Mode shopping experience featuring personal virtual try-ons and agentic checkout. Shoppers can upload a full-length photo to see how billions of apparel items drape and fit on their own bodies, or have AI automatically complete a purchase via Google Pay when prices drop.
Google just outlined its vision to turn Gemini into a universal AI assistant, aiming to evolve Gemini 2.5 Pro into a brain-like "world model." The company is integrating Project Astra's live capabilities and computer control into Gemini Live, while upgrading its Project Mariner browser agent to handle up to 10 tasks at once for Google AI Ultra subscribers in the U.S.
Google just rolled out major upgrades to its Gemini 2.5 lineup, introducing an experimental Deep Think reasoning mode for 2.5 Pro that tackles complex math and coding by exploring multiple hypotheses. The update also adds native multi-speaker audio dialogue, Project Mariner computer use in API, and native MCP support.
GitHub just introduced an autonomous coding agent for Copilot that works directly inside your existing workflow. Developers can assign a GitHub issue to Copilot, and it will spin up a GitHub Actions environment in the background, modify the codebase, and open a draft pull request. It also supports MCP servers and screenshot analysis.
Microsoft just baked its Phi-4-mini model directly into the Edge browser alongside new experimental AI APIs for prompting and text generation. Web developers can now tap local AI capabilities right on the client side without racking up backend costs or sacrificing privacy. The update also rolls out full-document PDF translation and brings Copilot agents directly to the Edge for Business sidebar.
Microsoft just launched Windows AI Foundry at Build 2025, overhauling its Copilot runtime for local inference across CPU, GPU, and NPU silicon. Windows 11 is also gaining native Model Context Protocol (MCP) support so AI agents can control desktop apps directly, alongside an open-sourced WSL.
Microsoft just introduced Microsoft Discovery at Build 2025, an agentic AI platform built to supercharge scientific research and industrial development. It combines a graph-based reasoning engine with specialized AI agents orchestrated by Copilot to tackle everything from hypotheses to HPC simulations. Microsoft researchers already used it to design a non-PFAS datacenter coolant prototype in roughly 200 hours.
Salesforce just rolled out Agentforce in Slack, shifting workplace AI from generic chatbots to task-specific "digital teammates." These agents can reason through multi-step workflows, search across team chats, and query company data, with pre-built enterprise templates arriving June 12.
NVIDIA just rallied major storage makers and ODMs around its AI Data Platform reference design to power agentic AI. By embedding GPUs and DPUs closer to massive enterprise data stores, the new hardware helps agents index, retrieve, and reason across private docs in real time without bottlenecking.
OpenAI just launched Codex, a cloud-based software engineering agent powered by codex-1, a specialized version of o3. It operates inside isolated cloud sandboxes to write features, execute test suites, and open GitHub PRs across multiple tasks in parallel. Why it matters: Developers can now offload entire background engineering workflows asynchronously rather than babysitting real-time completions.
Windsurf just rolled out its SWE-1 model family, trained specifically on end-to-end software engineering rather than simple code generation. The flagship SWE-1 rivals Claude 3.5 Sonnet in tool-call reasoning at a lower operating cost, while the lite and mini versions are free for all users.
Google DeepMind just unveiled AlphaEvolve, an evolutionary coding agent powered by Gemini that designs and optimizes complex algorithms. It has already reclaimed 0.7% of Google's global data center compute, accelerated key AI kernels by up to 32.5%, and improved decades-old math records like Strassen's matrix multiplication. Academic early access is currently in the works.
Perplexity just partnered with PayPal to bring agentic shopping directly into search queries. Rolling out this summer to U.S. Pro subscribers, users can research products, book travel, or buy tickets and pay instantly via PayPal or Venmo without leaving the chat interface.
Airbnb just rolled out a major app revamp allowing travelers to book local services and experiences without booking a stay. The company is also upgrading its customer support into a full AI concierge capable of answering queries and handling bookings directly. Why it matters: Airbnb wants to convert millions of window shoppers into active spenders.
