Google launches Gemini 4 Carbon, top model for cybersecurity and coding
Gemini 4 Carbon tops AI benchmarks, beats OpenAI and Anthropic in security tests, and launches at $2 / $10 per million tokens.
Google DeepMind released Gemini 4 Carbon on 28 Oct 2026, two weeks after its predecessor Argon.
Carbon leads the SEC‑bench Pro and ExploitGym security benchmarks, outperforming GPT‑6 Astra on SEC‑bench Pro and achieving an 88.2 % success rate in vulnerability discovery.
Google says the model can run multi‑day autonomous coding sessions, avoid prompt‑injection attacks and generate more transparent, honest outputs.
What it means for you
For AI users, Carbon offers a powerful option for security‑focused tasks and complex, long‑running code work, though access is currently restricted. Those seeking free or locally hosted models will still rely on community‑ranked options such as Kimi K3 or GLM‑5.3, as no single home machine can run Carbon.
Sources
- Google launches Gemini 4 Carbon with a ‘significant leap’ in cybersecurity, The Verge, 28 Oct 2026
- Google Deepmind releases Gemini 4 Carbon just two weeks after Argon with gains across the board, The Decoder, 28 Oct 2026
- Google DeepMind Introduces Gemini 4 Carbon, Its Top Model for Cyber Defense and Coding, MarkTechPost, 28 Oct 2026
Written automatically from the sources above, then checked by our code: every number and name must appear in them, and no sentence may copy a source. Spotted a mistake? Report it.
Which AI can you use for free today?
Every free AI model, ranked by blind human votes, with the memory and speed it needs on your own GPU.