Grok 4.6 matches GPT-5.6 Sol on composite evals with a 500K context window, at $2/$6 per million tokens—roughly half Sol's price—and is live in the xAI API, Cursor, Grok Build, OpenRouter, Vercel, and Cloudflare.
Why it matters: A frontier-tier coding and agent model at aggressive pricing changes the cost calculus for teams building agentic workflows on premium models—worth modeling against current spend.
Cowork sessions run directly in Chrome—history saved, skills and connectors active in-tab, tasks hand off to Claude desktop, web, and mobile; available on Max and Team plans now, rolling out to Pro.
Why it matters: Cowork embedded where browser work already happens removes context-switching from operator workflows that rely on live web data.
OpenAI rolled out sponsored placements to five new markets; free and Go-tier users see ads below relevant responses, labeled and separated from content, while paid subscribers stay ad-free.
Why it matters: Answer-engine advertising is no longer a US-only pilot—brands in these markets have a live new channel with intent-rich placement to evaluate and factor into strategy.