AI Product Updates: New Features and Releases

Oracle announced AI infrastructure deals, including bringing Google TPUs to Oracle Cloud in 2027 and four new OpenAI enterprise offerings.

A wrapper lets agents like Claude Code generate images, audio and text with an AI Studio key, no curl needed.

Google launched a stateless Gemini API CLI that lets existing coding agents generate media, analyze PDFs and YouTube URLs, transcribe audio, and run Deep Research.

A teardown of Antigravity 2.26 finds unreleased features including best-of-N battle mode, an experimental model library with Project Jarvis, sidecar plugins and built-in Chrome DevTools MCP.

A Thai morning digest covering seven updates including Meet live translation in 63 languages, M365 Copilot Projects, and the OpenAI Decisions API beta.

Google released a preview of gemini-api, a CLI that lets agents generate images, video, speech and music, analyze media and manage agents without hand-written curl scripts.

OpenAI's frontier models and Codex agentic workflows are now available in Google Cloud's Model Garden, letting enterprises use them where they already build.

Covers OpenAI's Orchestra Bio acquisition, ChatGPT Locked Chats and 12-hour prompt caching, plus news from Google, Mistral, Perplexity and more.

Google and Accenture launch a federal AI agent program using Gemini to automate government HR, finance and supply chain workflows, with Accenture handling deployment.

Google open-sourced a read-only Google Ads MCP, while Meta shipped an Ads MCP and CLI covering reporting, campaign management and A/B tests, with new campaigns paused by default.

Twelve AI product launches on Base, including API group-buying, agent payments, anonymous Gemini 4 access, and Dot Desktop's private browser and terminal.

Google DeepMind integrated AlphaFold, AlphaGenome and other science models into its Antigravity coding agent and launched a credits program for academic researchers.