Google unveils Gemini 4 Carbon, its strongest model yet, with staged access for cyber defenders
Two weeks after Argon, Google DeepMind has introduced Gemini 4 Carbon, which tops the Artificial Analysis Intelligence Index and posts large gains in vulnerability discovery and patching. Access starts with vetted cybersecurity teams through the Fairwind Program before a wider roll-out to developers and consumers.
Google DeepMind has announced Gemini 4 Carbon, describing it as the company's strongest model to date and the successor to Gemini 4 Argon, which launched only a fortnight earlier. Koray Kavukcuoglu, SVP of Google DeepMind and Google's Chief AI Architect, said Carbon builds directly on Argon's safety and security stack while pushing further on long-horizon work and cybersecurity.
On public leaderboards the model ranks first on the Artificial Analysis Intelligence Index v5.0 with a score of 59 and has climbed to fourth on WebDev Arena. Google says internal teams have let Carbon run for days on multi-step engineering tasks such as locating and patching defects across large codebases. The cybersecurity numbers are the headline: 88.2 per cent on SEC-bench Pro for vulnerability discovery, 54.8 per cent on ExploitGym and 71 per cent on CWE-bench for patching, with more than double Argon's performance on secure code generation.
Alignment metrics also improved. Google reports a 0.0 per cent attack success rate on the Gray Swan indirect prompt injection benchmark, reduced shortcut-taking on Terminal-Bench 4.0, and fewer unsupported claims and instances of overreach in multi-step tasks.
Precisely because of the offensive security capability, Google is staging the release. Carbon is initially available to a subset of cyber defenders and will expand to trusted partners through the Fairwind Program before reaching developers, businesses and consumers. Pricing matches Argon's introductory rate of $2 per million input tokens and $10 per million output tokens, rising to $4 and $20 after the introductory period.
Why it matters
The gap between model generations is now measured in weeks, and the leading labs are increasingly gating their most capable releases by use case rather than shipping to everyone at once. Enterprises should expect staggered availability to become the norm for frontier models with dual-use capabilities.
Indicops view
For Indian CERT-In-aligned security teams and critical-infrastructure operators, a model that can find and patch vulnerabilities at this level is a force multiplier, but only with strict scoping and human sign-off on every change. We recommend that organisations evaluating Carbon or Argon build an internal red-team harness first, so that capability can be measured against their own codebases rather than vendor benchmarks.
More in Models
Back to the newsroomGoogle launches Gemini 4 Hydrogen, a $1.50-per-million-token model for coding and agents
Google has released Gemini 4 Hydrogen, a faster and cheaper tier of its Gemini 4 family aimed at everyday coding and agentic work. It costs $1.50 per million input tokens and $7.50 per million output tokens across a 1M-token window, and is rolling out through the Gemini API, AI Studio and consumer products.
Alibaba opens a two-week preview of Qwen4-Max with a 1M-token context, but no benchmarks or weights
Alibaba Cloud has opened a time-limited preview of Qwen4-Max, its next flagship model, through the Token Plan only. It offers a 1M-token context and up to 393,216 output tokens, but the documentation lists no pricing, benchmarks or downloadable weights, and the model retires on 25 November.
Anthropic releases Claude Fable 5.2 and Mythos 5.2, 30 per cent faster and 20 per cent cheaper
Anthropic has released Claude Fable 5.2 for general use and Mythos 5.2 for vetted cyber and life-sciences organisations. The company says output is up to 30 per cent faster than the 5.1 generation, at $8 per million input tokens and $40 per million output tokens, with lower rates of safeguard blocks.