Gemini 4 Carbon arrives just in time to feel like yesterday's news in the fastest model cycle yet
The Decoder
Google has made its Gemini 4 Carbon AI model generally available, offering advanced reasoning and cybersecurity capabilities amid fierce competition.
Google has officially released Gemini 4 Carbon for paying developers and business customers following a two-week testing phase. Designed for intensive tasks like multi-step coding, debugging, and scientific research, Carbon complements the existing Gemini 4 Argon model by pushing intelligence limits, though it consumes more tokens during deep reasoning loops.
Alongside the model release, Google previewed an innovative "/swarm" feature in Antigravity that transforms a single prompt into a coordinated team of specialized AI agents. Carbon handles deep reasoning roles such as architecture design and auditing, while lighter models run in parallel workspaces.
In testing and benchmarks, Carbon achieved strong results, outperforming some competitors on metrics like CursorBench 4.0 while maintaining lower per-task costs than models such as Anthropic's Fable 5.2. Additionally, the model demonstrated cybersecurity utility by uncovering real zero-day vulnerabilities in under two hours during early-access testing.
Google offers Carbon through the Gemini API in Google AI Studio, Android Studio, and Gemini for Business at introductory pricing. The rapid release cycle highlights fierce competition in the AI sector as rivals frequently introduce outperforming models.
Why it matters: The release of Gemini 4 Carbon and its agent swarm capabilities highlights how major tech companies are rapidly advancing AI reasoning and multitasking features to stay competitive in a fast-moving model cycle.



