Playbooks · Coding tasks
Gemini 4 Carbon is here: $2 in, $10 out
Written November 11, 2026
Google's strongest model is now on the paid Gemini API at the same introductory price as Argon — and the price doubles on December 1, 2026. Carbon is built for the hardest tasks, not for everyday coding: Google still says Argon is the right pick for most software engineering and knowledge work.
The price, with the date attached
On the paid tier, per million tokens in US dollars, at the Standard tier: input is $2.00 through November 30, 2026, and $4.00 starting December 1, 2026. Output, including thinking tokens, is $10.00 through November 30 and $20.00 from December 1. Context caching is $0.10 through November 30 and $0.20 from December 1, and caching is implicit only: explicit caching is not supported. The rates are printed on Google's Gemini API pricing page, and the same page states that Carbon is not available on the Free tier — every token the model processes is counted and billed, including thinking tokens.
Batch and Flex run at half the Standard rates in both periods: $1.00 input and $5.00 output through November 30, $2.00 and $10.00 from December 1. Priority runs higher: $3.60 input and $18.00 output through November 30, $7.20 and $36.00 from December 1. A task that reads 100,000 uncached input tokens and writes 10,000 output tokens costs about $0.30 at the introductory Standard rates and about $0.60 at the December 1 rates. The arithmetic is exact at list prices; the token counts in real agent work are not, because thinking tokens are billed as output and are not visible in the answer.
Where Carbon can be reached
Carbon is available through the Gemini API on the paid tier in Google AI Studio and Android Studio, and through Gemini for business. Google AI Ultra subscribers can also use it in the Gemini app, in AI Mode in Search, and in Google Antigravity. It is also generally available in GitHub Copilot since November 10, for Pro+, Max, Business, and Enterprise users, billed at provider list pricing under usage-based billing, with the rollout reaching accounts gradually — if it is not in the model picker yet, check back rather than assuming it will never appear. Administrators on Business and Enterprise plans can manage it through the model policy in Copilot settings, and under default model enablement new models are switched on unless an administrator has turned that off.
When Carbon, and when Argon
Google's own line is the one to follow: Argon remains the ideal choice for the majority of software engineering and knowledge work, while Carbon is built for tasks that test the limits of a model — complex, multi-step coding and debugging, and open-ended scientific research and discovery. Carbon uses more tokens than Argon to explore thoroughly, reason deeply, and run extended verification loops. Picking Carbon for ordinary coding pays the same per-token price for longer thinking on work Argon would have finished sooner.
Carbon's strengths, as Google describes them, are long-horizon agentic work: multi-package migrations, cross-subsystem root-cause investigations, complex architectural redesigns, and web app development with build-and-verify loops that check both the code and the rendered interface. The model card and developer guide on Google's launch post carry the detail; the quality claims below are Google's own, not measurements of ours.
/swarm in Antigravity
Google is previewing /swarm in Antigravity alongside the Carbon launch: one prompt becomes a coordinated team of agents — a coordinator, parallel workers in isolated workspaces, and verifiers that test every change — tracked in real time. The Antigravity changelog describes it as coordinating multiple subagents over a message board to complete long-horizon tasks. Google's suggestion is to assign Carbon to the roles that demand the deepest reasoning, like architectural design, coordination, and auditing, and pair it with faster, lighter models for the parallel work. Carbon itself arrived in Antigravity on November 10, the same day as the API launch.
Carbon for cyber stays behind Fairwind
The cybersecurity version of Carbon is a separate door. Google describes Carbon for cyber as its strongest cybersecurity capability, leading on benchmarks for finding, exploiting, and patching vulnerabilities — but it is reached through the Fairwind Program, the defender program Carbon was first released to two weeks before the public launch. The launch post invites defenders to apply to join Fairwind for access. A builder who wants the cyber-tuned model should apply there, not assume the API model is the same configuration.
What Google claims about quality
Google says Carbon ranks first on CursorBench 4.0, Perplexity's WANDR, and Zapier's AutomationBench, at less than half the cost per task of the next best external model on those benchmarks. Google also partnered with Artificial Analysis on a new Multimodal Knowledge Agent benchmark, on which it says Carbon ranks second while costing more than six times less per task than the top model. Those are vendor claims from the launch post, stated here as Google's claims. An independent read of Carbon against other models is a separate piece of work, and no benchmark figure here should be treated as a measurement of ours.
Limits
- Prices here are Google's Gemini API list prices, paid tier, in US dollars per million tokens. What a subscription plan, a coding assistant, or a third-party gateway charges for the same model can differ, and those routes can carry different prices and rollout dates.
- The December 1 date is the date on the pricing page. If Google changes the page, the page governs, and any figure here should be reread against it before a budget is committed.
- No benchmark, quality, or capability claim is made here beyond Google's own statements. Do not pick Carbon as a default coding model on price alone — at the same per-token price as Argon, its longer thinking is the cost, not the rate.
Related playbooks
- Gemini 2.5 Pro and Flash retire December 1
- Gemini 4 Argon doubles in price on December 1
- Run Gemini 4 Hydrogen from the Copilot picker
- Gemini 4 Hydrogen is GA; there is no 4 Pro
Sources
- Gemini 4 Carbon is now available — Google Blog
- Gemini Developer API pricing — Google AI for Developers
- Changelog — Google Antigravity Docs
- Gemini 4 Carbon in GitHub Copilot — GitHub Changelog
Back to: Coding playbooks hub · Also: Gemini 4 Argon doubles in price on December 1 · Antigravity CLI 1.5.4: ACP Auto mode · MKA: the benchmark where Fable barely wins · Antigravity CLI 1.5.5: swarms in preview · Claude Fable 5.2 / Mythos 5.2: what changed