Google is rolling out its latest AI model, Gemini 4 Carbon, to the broad community following releasing it to a choose group of “cybersecurity defenders” two weeks ago. Google claims Carbon is its “strongest model” yet, and capable of competing alongside the extremely finest systems from Anthropic and OpenAI.
Google says Gemini 4 Carbon “pushes the frontier equal further” than Argon, referring to the company’s another frontier example that launched lone final month. The business stated Argon volition apt remain “the ideal choice for the bulk of application engineering and cognition work,” but that Carbon is “built for tasks that test the limits of a model’s intelligence,” specified as complex, multi-step coding and open-ended specialized research.
To emphasize Carbon’s usefulness for long-horizon activity involving AI agents, Google stated it is additionally previewing a “swarm” characteristic on its agentic betterment phase Antigravity, which volition let users rotate a sole immediate “into a coordinated squad of agents.”
Carbon is rolling out today via the Gemini API in Google AI Studio and Gemini for endeavor for all paying developers and enterprises, Google said in a blog post. It volition additionally be accessible to all Google AI Ultra subscribers in the Gemini app, the AI Mode in Search, and on Antigravity. It is not apparent whenever — or if — Carbon volition be accessible to lesser tier subscribers or liberated users.
As alongside the initiate of Argon, Google is pricing Carbon extremely competitively. The example is launching alongside an introductory cost of $2 per myriad input tokens and $10 per myriad output tokens, about a fifth the cost of OpenAI’s GPT-6 Astra and a fourth the disbursal of Anthropic’s Fable 5.2 model, which launched yesterday. Even following those expenses twice at the end November — whenever the introductory duration of lesser prices finishes — Carbon volition motionless be considerably cheaper than frontier rivals on a per-token basis, although this doesn’t necessarily translate to lesser expenses as models can use distinct numbers of tokens for the identical tasks.
Benchmarks cited by Google whenever it archetypal announced Carbon propose the example is among the most mighty on the market throughout most metrics, rearward Fable 5.2. It have to be noted that companies can and do cherry-pick benchmark results to display their tools in the finest imaginable light.
The release, hot on the heels of Argon, marks a decisive come back for Google at the extremely forefront of AI research. For months, frontier betterment looked akin a two-horse competition between Anthropic and OpenAI.
Follow topics and authors from this narrative to see additional akin this in your personalized homepage nourish and to obtain email updates.
