Google Debuts Gemini 4 Carbon AI Model With Advanced Cybersecurity And Coding Capabilities

ISTANBUL, Oct 29 – Google officially unveiled its newest frontier artificial intelligence model, Gemini 4 Carbon, on Wednesday, claiming major breakthroughs in software coding, autonomous agent execution, and cybersecurity defense.

The release comes just two weeks after the launch of its Gemini 4 Argon model, with Google attributing the accelerated delivery timeline to researchers and AI systems collaborating to speed up development.

In an official blog post, the tech giant highlighted that Gemini 4 Carbon claimed top spot on version 5.0 of the Artificial Analysis Intelligence Index with an overall score of 59.

Refinements built on feedback from the Argon release yielded performance jumps across full-stack web development, scientific task execution, and multimodal processing, including an 89-point gain to reach fourth place on the WebDev Arena leaderboard.

In internal safety and alignment testing, Carbon reduced unsupported claims and hallucination rates by nearly half compared to Argon, stating methodological caveats directly rather than fabricating evidence.

Google stated that Gemini 4 Carbon will initially roll out to cybersecurity defense teams this week before expanding availability to developers, enterprise clients, and general consumers.

Introductory API pricing will start at US$2 per 1 million input tokens and US$10 per 1 million output tokens (with a 95 percent discount for cached input tokens), before transitioning to standard rates of US$4 input and US$20 output per 1 million tokens.

Cybersecurity Metrics, Safety Refinements, and API Pricing

Key performance benchmarks, safety improvements, and pricing structures include:

• SEC-bench Pro Vulnerability Detection: Scores 88.2 percent in identifying system vulnerabilities.
• ExploitGym & Patching Scores: Registers 54.8 percent on ExploitGym exploitation testing and 71 percent on CWE-bench for code patching.
• Workflow Task Alignment: Reduces multi-step task overreach by 26 percent and under-service by 29 percent.
• Tiered API Pricing Model: Introductory rates set at US$2 per 1 million input tokens and US$10 per 1 million output tokens, before normalizing to US$4 input and US$20 output rates.

–NMT