Google announces Gemini 4 Carbon with top cybersecurity benchmarks

Google releases Gemini 4 Carbon, its strongest frontier model, featuring state-of-the-art cybersecurity and long-horizon agentic capabilities.
Carbon ranks #1 on the Artificial Analysis Intelligence Index v5.0 with a score of 59 and achieves 88.2% on SEC-bench Pro for vulnerability detection.
The model sets new state-of-the-art results on ExploitGym (54.8%) and CWE-bench (71.0%) for exploitation and patching, respectively.
Initial access is restricted to a subset of cyber defenders via the Fairwind Program, with broader availability planned after guardrail iteration.

Performance and Benchmarks

Jumps +89 elo points to rank #4 (1769) on WebDev Arena for full-stack web development.
Achieves a 0.0% attack success rate on the Gray Swan Indirect Prompt Injection Benchmark, resisting all evaluated attempts.
Reduces successful cheating on Terminal-Bench 4.0 to 0.0% while improving overall accuracy.

Cuts unsupported claims by approximately 2x compared to Gemini 4 Argon.
Reduces overreach by 26% and under-service by 29% in multi-step plan execution.
High-severity misalignment alerts dropped by almost 80% in internal live coding environments.

Introductory pricing is $2 per million input tokens and $10 per million output tokens.
Cached input tokens are priced at 95% off the standard input token price.
Post-introduction pricing will increase to $4 per million input tokens and $20 per million output tokens.