From the source
Google DeepMind released Gemini 4 Carbon, a frontier model it says is built for multi-day autonomous development workflows, two weeks after Gemini 4 Argon.
- Carbon tops the Artificial Analysis Intelligence Index v5 with 59 points, up from Argon's 50, and takes first place in four of the index's nine individual evaluations.
- Google says internal teams ran the model for multiple days straight, working through dead ends on its own and handling root-cause analysis and subtle bug fixes in large codebases.
- Because of its advanced cybersecurity capabilities, Carbon is going through the U.S. government's voluntary pre-testing process and will be available first to selected security experts starting this week, with access expanding through the Fairwind Program.
- Pricing stays the same as Argon.
- On Gray Swan's indirect prompt injection benchmark, Google says Carbon cut the attack success rate from 0.7 percent with Argon to 0.0 percent, though it notes prompt injection is not solved and the test appears saturated.