Gemini 4 Carbon is out, and its API price doubles after Nov 30
Quick answer: On Nov 10, 2026, Google made Gemini 4 Carbon, which it calls its strongest model, available to paying developers and enterprises through the API, and to Google AI Ultra subscribers in the Gemini app, AI Mode in Search and Google Antigravity. The API costs $2 per million input tokens and $10 per million output tokens until Nov 30, then $4 and $20. Google says Carbon ranks #1 on three agent benchmarks at more than 2X lower cost per task than the next best external model. Those are Google’s numbers, not independent tests. (Google blog, Nov 10)
What’s new
- Who gets it. Paying developers and enterprises via Google AI Studio and Gemini for business. It is “rolling out” to all Google AI Ultra subscribers in Google Antigravity, the Gemini app and AI Mode in Search. Free-tier users are not listed.
- Price. The introductory price matches Gemini 4 Argon: $2 per million input tokens, $10 per million output tokens. After Nov 30, 2026 it becomes $4 and $20.
- Carbon vs. Argon. Google says Argon “remains the ideal choice for the majority of software engineering and knowledge work.” Carbon is for the hardest tasks, and uses more tokens to “explore thoroughly, reason deeply, and run extended verification loops.”
- /swarm preview. In Google Antigravity,
/swarmturns one prompt into a coordinated team of agents you can watch in real time. It is a preview. - Cybersecurity. Google gave Carbon to security defenders through its Fairwind Program two weeks ago and says it leads on benchmarks for finding, exploiting and patching vulnerabilities. Defenders can apply to join Fairwind.
- Scientists. Google is expanding free six-month Antigravity access (Carbon and Argon) for current and future members of its 10,000-scientist program, subject to eligibility, capacity and selection.
What Google claims (and what that means)
- Benchmarks. Google says Carbon ranks #1 on CursorBench 4.0, Perplexity’s WANDR and Zapier’s AutomationBench, at more than 2X lower cost per task than the next best external model. It also says Carbon ranks #2 on a new Multimodal Knowledge Agent benchmark made with Artificial Analysis, at more than 6X lower average cost per task than the top model. These are results Google published about its own model; this brief has not checked them.
- Safety numbers. In Google’s simulated cyberattack tests, it says Carbon blocked 99% of attack attempts from its red-teaming agent, and its safeguards cut the per-attempt breach rate to 0.8% from 3.1% on Argon. Google also says outside testing found no new universal jailbreaks across 132 hours and over 13,000 requests, and that prompt-injection robustness “remains an area for continued improvement.”
- Signal proof. Google researchers say they used Carbon with a new tool called AlphaProof Code to prove key security properties of the code behind Signal’s cryptographic protocol “in weeks,” a task Google says historically takes expert teams years. This is Google’s description of its own research.
Should you care?
- If you build with AI: it costs the same as Argon until Nov 30, so you can test it on hard jobs before the price doubles. Check whether your tasks justify the extra tokens it uses.
- If you just use the Gemini app: Carbon is for Google AI Ultra subscribers for now.
- If you work in security: the Fairwind Program is the route to the cyber-tuned version.
FAQ
Is Gemini 4 Carbon free? Google’s post lists paid API access and Google AI Ultra. It does not say free-tier users get it.
What does it cost? $2 per million input tokens and $10 per million output tokens until Nov 30, 2026; $4 and $20 after.
How is Carbon different from Argon? Google says Carbon is for tasks that test the limits of a model’s intelligence and uses more tokens; Argon is recommended for most everyday engineering and knowledge work.
Are the benchmark results independent? The post reports rankings on named third-party benchmarks, but the results are published by Google. This brief did not verify them.
Sources
- Google, “Gemini 4 Carbon is now available” (Nov 10, 2026): https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-4-carbon-now-available/
— AltTab Dad