Google's Gemini 4 Argon targets the full robotics pipeline
An AI Frontline article republished by 36Kr says Google's Gemini 4 (Argon), described in Google's developer blog as a general physical-reasoning model for interacting with the physical world, scores 87% on the ERQA embodied-reasoning benchmark, ahead of Claude Opus 5.5 (82.3%) and GPT-6 Astra (81%). The piece says Argon can control real robots without robot-specific fine-tuning, generate runnable MuJoCo simulation code from images or video (Real2Sim), and automatically label and clean teleoperation data, with a context window of up to 1M tokens. It also cites community experiments on the Antigravity platform and Hacker News, and claims that fine-tuning a VLA model on 4,500-plus Argon-generated virtual trajectories raised real-robot success from 56% to 75% and cut average task time from 35 to 24 seconds. All figures come from Google or the article's own account and have not been independently reproduced.
If these capabilities hold up, labor-heavy robotics work such as simulation building, data labeling and evaluation could be largely automated by general models, reshaping how embodied-AI teams work.