Google ships Gemini 4 Carbon at $2 per million input tokens
Google's Gemini 4 Carbon is live for paying users at $2 per million input tokens — a fifth of GPT-6 Astra's cost — with a new agent 'swarm' feature on Antigravity.

- Gemini 4 Carbon launched today at $2 per million input tokens and $10 per million output tokens
- Introductory pricing is about one-fifth of OpenAI GPT-6 Astra and one-quarter of Anthropic Fable 5.2's cost
- Prices double at the end of November when the introductory period ends
- Carbon reaches AI Ultra subscribers in the Gemini app, AI Mode in Search, and Antigravity
- Google is previewing a 'swarm' feature that turns a single prompt into a coordinated team of agents
Google released Gemini 4 Carbon to the general public today, two weeks after giving the model to a select group of what it calls "cybersecurity defenders," and priced it aggressively: $2 per million input tokens and $10 per million output tokens.
That introductory price is roughly a fifth the cost of OpenAI's GPT-6 Astra and a quarter the cost of Anthropic's Fable 5.2, which launched yesterday. Even after the promotional rates double at the end of November, Carbon will remain significantly cheaper than frontier rivals on a per-token basis, Google said — though the company's blog post acknowledged that doesn't necessarily mean lower real-world costs, since models consume different numbers of tokens for the same tasks.
Google calls Carbon its "strongest model" yet, one it says can compete with the very best systems from Anthropic and OpenAI. The model, the company said, "pushes the frontier even further" than Argon, Google's other frontier model that launched just last month.
What is Carbon actually for?
Google positioned the two models for different jobs. Argon, it said, will likely remain "the ideal choice for the majority of software engineering and knowledge work." Carbon is "built for tasks that test the limits of a model's intelligence" — complex, multi-step coding and open-ended scientific research, the kinds of long-horizon problems where a model has to sustain effort across many steps.
That focus on long-running work is also behind the headline feature: a "swarm" capability Google is previewing on Antigravity, its agentic development platform. Swarm lets users turn a single prompt "into a coordinated team of agents" that work in parallel on a task.
Who gets access, and when?
Carbon is rolling out now via the Gemini API in Google AI Studio and Gemini for business, for all paying developers and enterprises. It's also available to Google AI Ultra subscribers in three places: the Gemini app, AI Mode in Search, and Antigravity.
Google did not say when — or whether — Carbon will reach lower-tier subscribers or free users. For search watchers, the AI Mode availability matters most: Ultra subscribers can now get Google's strongest model answering queries inside Search itself.
How strong is it, really?
Benchmarks Google cited when it first announced Carbon suggest the model is among the most powerful on the market across most metrics — but behind Anthropic's Fable 5.2. One caveat worth keeping in mind: companies can and do cherry-pick benchmark results to present their tools in the best light, so vendor-supplied numbers deserve skepticism until independent evaluations land.
Why this release matters
The timing is the story. Carbon arrives hot on the heels of Argon, and the back-to-back launches mark what looks like a decisive return for Google at the very forefront of AI research. For months, frontier development had settled into a two-horse race between Anthropic and OpenAI.
For the search industry, the open questions are practical. If Carbon's swarm feature makes agentic, multi-step work cheap and reliable, expect more automated query handling inside AI Mode — and more publisher anxiety about traffic. The price doubling at the end of November is the nearest fixed date to watch; after that, independent benchmark results and any sign of Carbon trickling down to non-Ultra users will tell us how far Google intends to push this model into mainstream Search.
via theverge.com (Source)
More from James Calloway
Show full bio
Market editor covering money, funding and market shifts around Search rankings and algorithm updates, generative engine optimization at SERP Journal.
220 articles