Gemini 4 Carbon is now available

Why it matters

Google has made Gemini 4 Carbon broadly available, framing it as its most capable model for work at the edge of AI reasoning: multi-step coding and debugging, open-ended scientific research, and cybersecurity. Carbon follows the earlier release of Gemini 4 Argon, which Google says remains the better fit for most software engineering and everyday knowledge work; Carbon uses more tokens to explore problems, reason in depth, and run extended verification loops. Google says Carbon ranks first on independent workflow benchmarks from Perplexity and Zapier, as well as a coding benchmark, while costing less per task than the next best external model. The model is accessible to developers and enterprises through Google AI Studio and Gemini for business, and to subscribers through Google Antigravity, the Gemini app, and AI Mode in Search. The announcement also highlights a new Google Antigravity feature, previewed under the name /swarm, that turns one prompt into a coordinated team of tracked agents, with Carbon assigned to the deepest reasoning roles. Google had previously tested Carbon with cybersecurity defenders via its Fairwind Program.