Gemini 4 Carbon is now available to developers, enterprises, scientists, and consumers — two weeks after its initial launch for cybersecurity professionals. Google confirmed the expanded rollout in a brief announcement published Monday evening, with no change to pricing or access tiers: only subscribers to the AI Ultra plan get it in the Gemini app.
Everyone else must use AI Studio or Android Studio to integrate it into apps or workflows. No new hardware, no new device partnerships, no regional restrictions, just wider API and app-level access, effective immediately. It’s not a standalone product. It’s a model, one that Google says outperforms both its own Gemini 4 Argon and rival models from OpenAI and Anthropic across knowledge work, agentic coding, math, science, and cybersecurity benchmarks.
It also scores 0% on the Gray Swan IPI prompt-injection test, a measure of resistance to adversarial input manipulation. That matters most to enterprise security teams and developers building production-grade AI tools, not to someone asking Gemini for dinner ideas. Spatial reasoning and physics simulation improvements are cited specifically for 3D game engines and creative coding.
Be the first to comment
Join the argument. No password, just your email or a passkey.