Gemini 4 Argon

Google has launched Gemini Robotics 2.5, its most capable vision-language-action model to date, delivering gains in dexterity, reliability and safety across more than 200 tested tasks. In a real-world proof of concept on a Mercedes-Benz production line, it drove Apptronik's Apollo 2 humanoid to pick and pack auto parts with a 96% success rate, improving success by 30-100% and speed by 20-30% over prior baselines. The model is now opening to Google's robotics partners.

Google has opened access to Gemini 4 Carbon, its most capable reasoning model to date, targeting complex software engineering, multi-step agentic tasks, scientific research and cybersecurity. It offers a unified context window of roughly 1.05 million tokens, accepts text, image, video, audio and PDF inputs, and provides three reasoning-effort tiers; it is available to paid developers via the Gemini API and AI Studio at an introductory price of $2 per million input tokens and $10 per million output tokens.

Google has made Gemini 4 Carbon generally available after a two-week testing phase, positioning it for demanding work such as multi-step coding, debugging and scientific research, complementing Gemini 4 Argon at the cost of higher token consumption. Google also previewed the /swarm feature in Antigravity, which turns a single prompt into a coordinated team of agents. Arriving amid rapid rival releases, the launch already feels late in the current model cycle.

Google has opened Gemini 4 Carbon, its strongest model, to all paying developers and Google AI Ultra subscribers, ending a two-week restriction to cybersecurity defenders in its Fairwind Program. Carbon keeps Gemini 4 Argon's introductory pricing of $2/$10 per 1M input/output tokens but spends more tokens on reasoning and verification, ranking first on CursorBench 4.0, WANDR and AutomationBench at less than half the per-task cost of the next-best outside model.

Google DeepMind unveiled Gemini Robotics 2.5, a vision-language-action model it says outperforms its own prior baselines by 30 to 100 percent depending on task type. On active Mercedes-Benz production lines in Hungary and Germany, Apptronik's humanoid Apollo 2 sorted metal parts with a reported 96 percent success rate. DeepMind also uses Gemini 4 Argon as a simulation engine for training and open-sourced the ASIMOV-Jailbreak benchmark to test robotics model safety.

Google has made Gemini 4 Carbon generally available to developers, enterprises, scientists, and consumers, positioning it as its strongest frontier model for cybersecurity capabilities and long-horizon agentic workflows. Carbon sits above Gemini 4 Argon and targets extreme tasks such as complex multi-step coding, debugging, and open-ended scientific research, while Argon remains the default for most software engineering and knowledge work. Carbon was previously released to select cybersecurity defenders through the Fairwind Program.

Google has made Gemini 4 Carbon generally available to developers, enterprises, scientists, and consumers, positioning it as its strongest frontier model with leading performance in cybersecurity and long-horizon agentic workflows. Previously released to defenders via the Fairwind Program, Carbon targets complex multi-step coding, debugging, and open-ended scientific discovery, while Gemini 4 Argon remains the default for most software engineering and knowledge work.

Google has opened Gemini 4 Carbon to the general public just two weeks after a limited release to cybersecurity defenders, calling it its strongest model yet and positioning it for complex multi-step coding and open-ended scientific research. It also previewed a swarm feature on its Antigravity agent development platform, complementing last month's Argon release.

Artificial Analysis reports that Claude Fable 5.2 (Max) now ranks first on AA-3D, its benchmark for interactive 3D generation, scoring 1271 Elo, 67 points ahead of Claude Sonnet 5.5 and Opus 5.5, with a confidence interval that overlaps no other model. It gains 132 points over its predecessor Fable 5.1 and beats the top OpenAI and Google entries, GPT-6 Astra (1168) and Gemini 4 Argon (1147).

A new Apple paper introduces AMP1, a harness that lets general-purpose frontier models such as Claude Opus 5.5 and Gemini 4 Argon control a robot arm in a closed loop using camera views, code execution and direct motion commands, with no demonstrations or task-specific training. It achieves 85.6% zero-shot success on RoboLab-120 versus 49.2% for the best VLA and 42.9% for the best world-action model, and succeeds in 104 of 108 trials on a real Franka arm, though the robot idles much of each episode waiting on model inference.