Google confirms Carbon’s restricted pre-release, access cohort, vendor benchmarks and future pricing; no public API or weights are available.
Google restricts Gemini 4 Carbon to selected cyber defenders
2026-10-29Verified announcementWATCH
Google introduced Gemini 4 Carbon as a cybersecurity-focused frontier model, but access is limited to a subset of defenders and will expand through a phased trusted-partner program. It is not available to general developers, businesses or consumers. Google reports gains on several cyber benchmarks and announced introductory API pricing for a later launch; those results are vendor-run and the public launch date is not fixed.
Official · ReferenceGemini 4 Carbon: our next frontier modelEvaluationGemini 4 Carbon on Artificial Analysis Intelligence Index v5
Availability: pre-release; selected cyber defenders only; phased access through the Fairwind trusted-partner program
What this means for you
For security teams, Carbon signals stronger long-horizon cyber automation, but restricted access prevents reproducible evaluation and general integration. Treat price and benchmark claims as planning inputs, not deployed capability.
Evidence & related material
Artificial Analysis independently reports Carbon at 59 and first on Intelligence Index v5, and second on its Coding Agent Index; this does not change the restricted pre-release access.
Technical details & scores
| Availability | pre-release; selected cyber defenders only; phased access through the Fairwind trusted-partner program |
|---|---|
| General Access | not available to developers, businesses or consumers |
| Vendor Benchmarks | AA Index 59; SEC-bench Pro 88.2%; ExploitGym 54.8%; CWE-bench 71% |
| Announced Launch Price | $2/M input tokens and $10/M output tokens; cached input 95% off; eventual price announced as $4/$20 |
| Evidence Limit | benchmarks and safety evaluations are Google-reported; no public API or weights |
| Time Basis | Google RSS published 2026-10-28T17:00:00Z |
| Independent Evaluation | Artificial Analysis Intelligence Index v5: 59 (#1); Coding Agent Index: second to Claude Opus 5.5; evaluation article is date-only and model access remains restricted |
Technical impact 5
Engineering value 3
Adoption signal 2
Maturity 2
Verification cost 4
Risk 5