Google executive announces Gemini 4 Argon rollout
Mirrokni says Argon is rolling out to AI Ultra and Pro subscribers, AntiGravity and developers, and stands out on multi-agent loops, math proofs and science tasks.
Gemini 4 Argon is finally rolling out to Google AI Ultra & Pro subscribers in the Gemini App, Ultra subscribers in AntiGravity, and to developers and enterprises! 🚀 We have been using this model across several complex engineering, math, and science tasks, and we are very pleased with its performance. Koray’s launch announcement highlights how Gemini 4 Argon scales beautifully with multi-agent loops . For instance, we've applied it directly to our AntiGravity agents using the /teamwork-preview framework. By partitioning agent contexts and running continuous verification loops, we've successfully optimized Google fleet memory, acheived new results in science, and formally verified complex mathematical proofs like the Courtade–Kumar (CK) conjecture in Lean and more new open problems; We will announce more, so stay tuned! This is a true step-change in Gemini models. See the launch instructions in Sundar’s post, and try it out today in the Gemini App, Gemini Enterprise, and AntiGravity! 💡👇 https://lnkd.in/gFimrfvX
This post quotes another post: “Gemini 4 Argon launches for hardest coding and knowledge work”