Google launched Gemini 4 Argon on October 13, 2026, claiming the model outperforms GPT-6 Astra, Claude Fable 5.1, and Claude Opus 5.5 across benchmarks in coding, cybersecurity, science, finance, and law2. Alongside the general release, Google DeepMind published a robotics-specific blog post detailing how Argon applies across the robotics stack, from robot control and real-to-sim reconstruction to physical understanding and research self-improvement1.
Robotics capabilities
On the Embodied Reasoning (ERQA) benchmark, which tests a model's ability to understand and solve physical, spatial problems in 3D environments, Argon scores 87% using pure reasoning, outperforming every other frontier model, according to Google DeepMind. The blog post describes four application areas: robot control, including mobile manipulation with MuJoCo practice, zero-shot robot control, and closed-loop steering of video world models; Real2Sim reconstruction of physically calibrated simulation twins from robot camera feeds, egocentric human video, or hand-drawn sketches; physical understanding via agentic multimodal hindsight labeling, dense value functions, and progress estimation; and research self-improvement through simulation and iterative hardware refinement. One demonstration shows Argon controlling a Boston Dynamics Spot quadruped with 12-DoF legs and a 6-DoF arm to learn a new skill: stepping on a foot-pedal trash bin to throw away soda cans.
Broader model release
Google says Argon delivers "frontier reasoning and improved performance across complex workflows" while cutting down on hedging, repetition, and clichés in chatbot responses. The company says it is already using Argon internally to find and patch cybersecurity vulnerabilities, halve the computing power needed to run YouTube's comments section, and debug the drone physics simulator for Alphabet's drone delivery company Wing. The model is rolling out to Google's higher-tier Pro and Ultra subscribers in the Gemini app and AI mode in Search, as well as to Ultra subscribers on its agentic development platform Antigravity.
ANALYSIS The simultaneous general-purpose and robotics-specific framing positions Gemini 4 Argon as both a frontier reasoning model competing with Anthropic and OpenAI and a dedicated embodied-AI system, a dual role none of its named rivals have claimed with a single model release.
The launch arrives on the same day Anthropic published research warning that AI self-replication on consumer devices could be months away[1], and the same week Meta fielded its first frontier models with Muse Wonder Full[2]. Google DeepMind describes Argon as "a step change in physical understanding, code generation, and visual-semantic reasoning" that accelerates progress toward what it calls physical AGI.