AI’s new battle: defense, agents, and 1M context

AI’s biggest moves right now aren’t just model launches. They’re going into the Pentagon, agent workflows, and everyday apps fast. Here are the latest shifts that matter most today.^[1][2][3]

Pentagon deal: DoD signed AI agreements with OpenAI, Google, Microsoft, AWS, Nvidia, SpaceX, and Reflection for classified networks, while Anthropic was left out.^[1] That could reshape who gets the government AI money.

OpenAI’s Codex just got much bigger: computer use, web workflows, image generation, memory, automations, and deeper dev tools are rolling out together.^[2] The big question: are we moving from chat to true AI coworkers?

Anthropic says Claude Opus 4.6 now has a 1M token context window in beta, plus stronger coding, research, and long-task performance.^[3] That means whole books, codebases, and long docs are finally in reach.

Google is pushing Gemini beyond chat too: it is rolling into cars, creating downloadable files, and adding more proactive help.^[16][24] If AI can answer and act in the car, search behavior may change fast.