Requests naming gemini-2.5-pro will be served by gemini-3.8-flash, and gemini-2.5-flash by gemini-3.5-flash-lite, billed at the new models' rates. Both moves bring breaking API changes, and 3.8 Flash's price doubles on 1 January.
Google opens Gemini 4 Carbon to paying developers at $2 and $10
EditorialBy TrustList Editorial
The introductory price per million input and output tokens ends 30 November 2026, when it doubles to $4 and $20. Google says Carbon uses more tokens than Argon.
- United States of America
- California
- Mountain View, California
- Artificial Intelligence - AI
- +3 more
About Google opens Gemini 4 Carbon to paying developers at $2 and $10
Google opens Gemini 4 Carbon to paying developers at $2 and $10
10 November 2026: Google has made Gemini 4 Carbon, which it calls its strongest model, generally available through the Gemini API at an introductory $2 per million input tokens and $10 per million output tokens. The introductory price ends on 30 November 2026; Google's pricing page lists $4 and $20 from 1 December.
Who can use it now
Koray Kavukcuoglu, Google's chief AI architect, announced on the company blog that Carbon is available to developers, enterprises, scientists and consumers. Through the API it is open to all paying developers and enterprises in Google AI Studio and in Gemini for business. Google's pricing page says Carbon is not available on the free tier.
Google AI Ultra subscribers are getting it in Google Antigravity, the Gemini app and AI Mode in Search. Cybersecurity defenders can apply to the Fairwind Program, where Google released Carbon two weeks earlier. Scientists in Google's free six-month Antigravity programme get access to both Carbon and the earlier Gemini 4 Argon.
The prices as Google lists them
Google says Carbon carries the same introductory price as Gemini 4 Argon. The pricing page gives these figures per million tokens on the paid tier:
- Standard input: $2.00 through 30 November, $4.00 from 1 December
- Standard output, including thinking tokens: $10.00 through 30 November, $20.00 from 1 December
- Context caching: $0.10 through 30 November, $0.20 from 1 December, with implicit caching only
- Batch input and output: $1.00 and $5.00 through 30 November, then $2.00 and $10.00
Grounding with Google Search includes 5,000 free requests a month, shared across Gemini 3 and newer models, then $14 per 1,000 requests. Google's footnote says the post-introductory price applies to the Carbon model intended for public access.
Same price as Argon, more tokens per answer
Google says Argon remains the right choice for most software engineering and knowledge work, and positions Carbon for tasks that test the limits of a model, such as multi-step coding and debugging and open-ended scientific research. It also says Carbon uses more tokens than Argon because it explores thoroughly, reasons deeply and runs extended verification loops.
That matters for budgeting, because a bill depends on tokens consumed as well as the price of each. Google's own comparison is in cost per task rather than per token.
Benchmark claims, as Google reports them
Google says Carbon ranks first on CursorBench 4.0, Perplexity's WANDR and Zapier's AutomationBench, at more than twice as low a cost per task as the next best external model. On the Multimodal Knowledge Agent benchmark, built with Artificial Analysis, it says Carbon ranks second while costing more than six times less per task than the top-performing model. These are Google's reported results.
Google is also previewing a /swarm feature in Antigravity, which turns one prompt into a coordinated team of agents that developers can track in real time.
Sources
- Google: Gemini 4 Carbon is now available, 10 November 2026, 10 November 2026
- Google AI for Developers: Gemini Developer API pricing, 10 November 2026
Categories & features
- United States of America
- California
- Mountain View, California
- Artificial Intelligence - AI
- Machine Learning
- Cloud
- API
TrustList Weekly
The week in software and IT, in one email
The news that matters to buyers, new rankings and our own research. Every Thursday, free, and easy to leave.
You may be interested in
Standard input rises from $2 to $4 per million tokens and output from $10 to $20; batch, flex and priority rates double on the same day, and the model stays off the free tier.
Video output falls 30% to $12 per million tokens on the new generally available model, against $17.50 on 1.1. Input and text output prices stay the same, and paid-tier rate limits double.
From 1 December, Gemini API calls to gemini-2.5-pro go to gemini-3.8-flash and calls to gemini-2.5-flash go to gemini-3.5-flash-lite, billed at the new rates. Several request parameters stop working. New projects now store requests by default.
The Evals dashboard and API, the v1/prompts API and Agent Builder all shut down on 30 November. OpenAI points Evals users to Promptfoo and Agent Builder users to its Agents SDK; older image models go on 1 December.
The release gives no gigawatt figure or date for the expansion, and says OpenAI trained GPT-6 Astra on an OCI Supercluster in Abilene.
It follows a $400m Upper90 debt facility. The cloud runs Cerebras, SambaNova and Positron AI chips, with first major deployments due in Q1 2027.
The common stock offering was raised from $22.5bn and is sold through forward sale agreements, so Meta gets no proceeds until settlement. Nvidia says it will invest $3bn.
More on TrustList
Everything here links back to the same verified catalogue. Pick your next stop.
- More United States of AmericaThe ranking for this subject
- CompaniesAgencies, consultancies and IT service providers, ranked by verified reviews.
- ProductsSoftware and SaaS with pricing, features, integrations and alternatives.
- AwardsAnnual recognition decided by verified reviews and an independent jury.
- LaunchesNew products and releases, voted up by the community every day.
- AI ModelsBenchmark scores and community ratings for every major model.
- RequestsBuyers describe what they need; vendors respond directly.
- PeopleReviewers, authors and makers with public profiles.
- ComparePut up to four listings side by side before you shortlist.