Updated
Updated · Ars Technica · Oct 28
Google Unveils Gemini 4 Carbon, Doubling Secure Code Output Over Argon
Updated
Updated · Ars Technica · Oct 28

Google Unveils Gemini 4 Carbon, Doubling Secure Code Output Over Argon

3 articles · Updated · Ars Technica · Oct 28

Summary

  • Two weeks after Gemini 4 Argon, Google introduced Gemini 4 Carbon as a stronger frontier model but is keeping it in a staged release rather than opening broad access immediately.
  • Carbon’s biggest claimed advance is cybersecurity: Google says it is twice as likely as Argon to produce vulnerability-free code and delivers a significant leap in detecting, validating and fixing flaws.
  • Google’s Chrome team said Carbon is 10 times better at generating high-quality patches than non-Google models, while the model also gained 89 points on WebDev Arena to reach No. 4.
  • Across Google’s benchmarks, Carbon edges Argon in every test and beats GPT-6 Astra, Claude Opus 5.5 and Claude Fable 5.1 in most categories, though the margins are only a few points.
  • The launch extends Google’s rapid post-pause model cadence, but the company has not clarified how distinct Carbon is from Argon beyond positioning it as a new model.

Insights

Can Google’s Gemini 4 Carbon really patch real-world vulnerabilities safely, or are benchmark wins hiding the hardest cybersecurity failures?
Why is Google limiting Gemini 4 Carbon to select defenders if it already claims top scores in coding, patching, and prompt-injection resistance?