Releases

Perplexity introduced local and hybrid execution that moves selected Computer workloads onto users' machines.

Microsoft AI introduces Frontier Tuning, using reinforcement-learning environments to adapt MAI models to organization-specific workflows.

Microsoft introduces Agent 365 as the trust, management and observability layer for enterprise AI agents.

Microsoft IQ becomes generally available across GitHub Copilot, Microsoft Foundry and Copilot Studio as an enterprise context layer for agents.

Microsoft introduces an integrated enterprise agent platform spanning GitHub, Foundry, Microsoft IQ, Copilot Studio, security and governance.

Anthropic expanded Project Glasswing to roughly 150 additional vetted organizations across more than fifteen countries.

OpenAI expanded Codex with reusable plugins and a Sites workflow for building and publishing complete web experiences.

Baidu released an enterprise edition of its general-purpose productivity agent.

SpaceXAI added the Composer 2.5 model option for long-running tasks and complex instruction following in Grok Build.

CoreWeave announced an industry-first bring-up and validation milestone for the NVIDIA Vera Rubin NVL72 platform.

NVIDIA announced the Vera Rubin platform had ramped into full production for agentic AI factories.

NVIDIA announced an open humanoid reference design to accelerate physical AI research and development.

Oracle expanded guardrails to moderate image inputs and support version pinning.

Perplexity brought Computer workflows into Microsoft Excel, Word, PowerPoint and Outlook.

Mistral AI expanded Vibe with a unified agent experience across coding and software-development workflows.

Mistral AI launched Search Toolkit for building grounded search, retrieval and research applications.

Mistral AI introduced Physics AI for industrial engineering, simulation and scientific workloads.

SpaceXAI launched Grok Build, an agentic coding product running from the terminal, for SuperGrok and X Premium+ subscribers.

Mistral AI Studio added Model Context Protocol connectors for integrating enterprise systems and tools.

Le Chat added remote agents and a Work mode for enterprise knowledge and task execution.

Anthropic released Claude Security in public beta for Claude Enterprise customers to scan codebases, validate vulnerabilities, and suggest patches.

Google expands Antigravity with new capabilities and integration into Gemini Enterprise Agent Platform.

Perplexity added data analysis, coding and visualization workflows for business teams to Computer.

CoreWeave launched secure isolated sandboxes for reinforcement learning, agent tool use, and model evaluation at scale.

Scale completed SWE Atlas by adding Test Writing and Refactoring alongside Codebase QnA, producing a 284-task software-engineering evaluation suite.