Now Reading
Google Launches Gemini 4 Carbon for Cybersecurity

Google Launches Gemini 4 Carbon for Cybersecurity

Google Gemini 4 Carbon AI Model Visual

Google has launched Gemini 4 Carbon, a new AI model designed for advanced coding, software security and complex engineering tasks. Announced on October 28, 2026, the model strengthens Google’s competition with OpenAI and Anthropic in the race to build more capable AI systems.

Unlike general-purpose assistants, Gemini 4 Carbon places particular emphasis on cybersecurity defence. It can help security teams identify software vulnerabilities, assess potential exploits and develop patches. Moreover, Google says the model can sustain complex, multi-step engineering work over extended periods.

The launch follows Gemini 4 Argon, which Google introduced on September 30. Consequently, Carbon represents another step in the company’s effort to improve coding performance and strengthen its position in the frontier AI market.

Cybersecurity Performance and Benchmark Results

Google reports strong results for Carbon across three cybersecurity benchmarks covering vulnerability detection, exploit validation and software remediation.

  • SEC-bench Pro: Carbon scored 88.2% on tests designed to identify vulnerabilities in real-world software.

  • ExploitGym: The model achieved 54.8% on tasks that assess whether vulnerabilities can be exploited.

  • CWE-bench: Carbon scored 71% on tests involving developing software fixes.

Together, these results address several stages of defensive security work. First, teams need to locate weaknesses. Next, they must determine which flaws pose practical risks. Finally, they need to fix the underlying causes rather than rely on temporary workarounds.

Google also reports that Carbon generates secure code at more than twice Argon’s success rate on internal benchmarks. In addition, the company says its Chrome team observed substantial improvements in security patch quality. These findings come from Google’s own testing, so independent evaluation remains important.

Designed for Complex, Long-Running AI Tasks

Gemini 4 Carbon also targets engineering projects that require sustained reasoning and multiple steps. According to Google, internal teams have seen it work for days while investigating difficult bugs across large codebases and low-level systems.

Beyond cybersecurity, the model improves coding, scientific tasks, agentic workflows and full-stack web development. Furthermore, Google reports an 89-point gain on WebDev Arena, where Carbon reached fourth place with a score of 1,769.

The company has also strengthened safeguards against prompt injection and unreliable outputs. Google reports that Carbon reduced unsupported claims by approximately half compared with Argon. Meanwhile, its internal testing showed a 0% attack success rate on the evaluated indirect prompt-injection benchmark. However, these results reflect specific test conditions and do not guarantee complete protection in every deployment.

Limited Availability and Pricing

Google is initially offering Gemini 4 Carbon to selected cybersecurity defenders through its Fairwind programme. It plans to extend access gradually to trusted partners while refining safety measures. However, Google has not announced a firm date for general availability to developers, businesses and consumers.

The model will launch at the same introductory API rates as Gemini 4 Argon.

See Also
ICLDEC and RIT Dubai Partnership Signing

  • Input tokens: $2 per million tokens.

  • Output tokens: $10 per million tokens.

  • Cached input tokens: 95% below the standard input price.

After the introductory period, Google plans to raise prices to $4 per million input tokens and $20 per million output tokens.

The launch comes as technology companies invest in AI tools for software development and cyber defence. For security teams, Carbon could help speed up vulnerability research and patch development. Nevertheless, organisations will need to validate its findings, review generated code and maintain human oversight.

Ultimately, Gemini 4 Carbon signals Google’s growing focus on AI systems that can support demanding technical work. Its broader impact will depend on independent performance testing, real-world security outcomes and how safely Google expands access.

View Comments (0)

Leave a Reply

Your email address will not be published.

© 2024 The Technology Express. All Rights Reserved.