Google has launched Gemini 4 Argon and made it globally available today to developers and enterprises through the API in AI Studio and the Gemini platform. Paid access is also rolling out for Google AI Ultra subscribers in Google Antigravity, while Google AI Ultra and Pro subscribers can use it in the Gemini app and AI Mode in Search.
The timing matters because Argon is not a niche release tucked inside one product. It is now spread across developer, enterprise and consumer-facing surfaces at once, which makes it the clearest sign yet that Google wants the model used in daily work as well as in search and chat. The launch price is $2 per million input tokens and $10 per million output tokens, with cached input tokens 95% off.
Google describes Gemini 4 Argon as a frontier model built for deep work in complex, real-world tasks across coding, finance, legal, cybersecurity and science. It also says Argon ranks No. 1 on the Vals Index, which measures model performance across finance, coding, legal and tax work. But the company has not published the benchmark scores or the margin over other frontier models, so the claim lands with the ranking attached and little else to inspect.
Even so, Google is pointing to early use as evidence that the launch is not just marketing copy. The company says software engineers have spent the past few weeks shipping high-quality code faster than ever using Argon on hard engineering problems. In Google Antigravity, it says Argon helped teams increase agentic code submissions by more than 35% and cut code rollbacks by nearly a third.
Google is also using demos to show where it thinks the model fits. Argon built a detailed model of a Pagoda using multi-agent visual critique loops in Antigravity, sculpted a multi-building temple complex in Blender, and turned pictures of a real desk into a playable miniature racing game. In that example, the desk objects, room and race track were modeled and collision-checked using Blender MCP, then textured with Nano Banana and wired with physics, custom shaders and car engine audio from Gemini Omni Flash in Three.js.
Those demonstrations are designed to show breadth, but they also raise a practical question about how far the rollout will go on the current terms. Google has said the introductory rates are $2 per million input tokens and $10 per million output tokens, yet it has not said when those prices will change. For developers and enterprises deciding whether to build around Argon now, that unanswered detail matters almost as much as the launch itself.
For now, the model is out, the products are live and the price is set. What comes next is whether Google keeps the same terms long enough for teams to make Argon part of their regular workflow, or moves the pricing before the first wave of adopters has settled in.
