Google · 2026-10-28 · seismic
Gemini 4 Carbon — Google's new frontier model, first for cyber defenders
Gemini 4 Carbon is Google's new frontier model for cyber defence and long agentic coding. It ranks #1 on the Artificial Analysis Intelligence Index v5.0 with 59, and goes first to a subset of cyber defenders, not to the public.

Google's Gemini 4 Carbon tops the Artificial Analysis index and sets new security scores, but only cyber defenders get it at launch.
Key specs
| Aa intelligence index v5.0 | 59 |
|---|---|
| Sec bench pro | 88.2% |
Quick facts
| Maker | Google DeepMind |
|---|---|
| Announced | 28 October 2026 |
| Availability | Subset of cyber defenders, then Fairwind Program partners |
| Public API | Not yet; no date given |
| WebDev Arena | #4, 1769 Elo (+89) |
| Intro price | $2 in / $10 out per 1M tokens |
Pricing
| Input · introductory, then $4.00 | $2.00 / 1M tokens |
|---|---|
| Cached input · of the input price | 95% off |
| Output · introductory, then $20.00 | $10.00 / 1M tokens |
What is it?
Gemini 4 Carbon is the second Gemini 4 model, announced on 28 October 2026, two weeks after Gemini 4 Argon. Google calls it its strongest model, with state-of-the-art cybersecurity and long-horizon capabilities for jobs like root-causing and patching bugs across large codebases.
How does it work?
On security tests, Carbon scores 88.2% on SEC-bench Pro for finding vulnerabilities, 54.8% on ExploitGym for exploitation and 71.0% on CWE-bench for patching; Google says all three ran without guardrails. Google also reports 0.0% successful cheating on Terminal-Bench 4.0 and 0.0% attack success on the Gray Swan prompt-injection test, against 0.7% for Argon.
Why does it matter?
A model that finds and exploits bugs this well helps defenders and attackers alike, so Google is giving security teams a head start before any wide release. For everyone else, the new #1 spot on the Artificial Analysis index raises the bar that OpenAI and Anthropic models are measured against.
Who is it for?
security teams, critical-infrastructure operators, frontier-model watchers
Frequently asked questions
- Can developers use Gemini 4 Carbon on the Gemini API?
- Not at launch. Google says Gemini 4 Carbon goes first to a subset of cyber defenders and then, step by step, to more trusted partners through its Fairwind Program. Google plans to bring Gemini 4 Carbon to developers, businesses and consumers after gathering feedback and improving guardrails, but it did not give a date.
- How much will Gemini 4 Carbon cost?
- Google lists an introductory price for Gemini 4 Carbon of $2 per million input tokens and $10 per million output tokens, with cached input tokens 95% off the input price. After the introductory period the price rises to $4 per million input tokens and $20 per million output tokens, the same schedule Google used for Gemini 4 Argon.
- How does Gemini 4 Carbon compare with Gemini 4 Argon?
- Google reports that Gemini 4 Carbon has more than 2x the secure-code success rate of Gemini 4 Argon on internal benchmarks, about half as many unsupported claims, and almost 80% fewer high-severity misalignment alerts in internal coding environments. On the Gray Swan indirect prompt-injection benchmark, Carbon has 0.0% attack success against Argon's 0.7%.
- Is Gemini 4 Carbon the top model on the Artificial Analysis index?
- Yes. Google says Gemini 4 Carbon ranks #1 on version 5.0 of the Artificial Analysis Intelligence Index with a score of 59. Version 5.0 is the rebuilt index that Artificial Analysis published on 27 October 2026, where private test sets carry 60% of the weight. On WebDev Arena, Gemini 4 Carbon sits at #4 with 1769 Elo.