🤖 Google releases Gemini 4 Argon, with a physical reasoning score of 87% Google releases a new generation of multi-modal reasoning model Gemini 4 Argon, which strengthens physical understanding, code generation and visual semantic reasoning

🤖 Google releases Gemini 4 Argon, with a physical reasoning score of 87%. Google releases a new generation of multi-modal reasoning model Gemini 4 Argon, which strengthens physical understanding, code generation and visual semantic reasoning. On the embodied reasoning benchmark ERQA, it scores 87% for pure reasoning, leading other cutting-edge models. Argon can control the Boston Dynamics Spot quadruped robot to complete tasks such as fetching and throwing objects, and can generate simulation data, increasing the success rate of real robots from 56% to 75% and reducing the time from 35 seconds to 24 seconds; it is now available through the Google-genai SDK. Pricing is $2 per million input tokens, per million output…