Google Pushes Gemini 4 Argon Into the Market With Steep Discounts

Google has put Gemini 4 Argon on sale worldwide, moving it beyond the small circle of cybersecurity partners that first got access at the end of September. The model is now available through the Gemini API, in Google’s business products, and to paying consumers on Google AI Pro and Ultra. Free users are out. Google has also stripped the Flash and Pro models from them recently, which tells you exactly where the company wants the value to sit.

The push is broad. Developers can use Argon through the paid tier of the Gemini API. Ultra subscribers can try it in Antigravity, AI Studio and Android Studio. Businesses get it in Gemini for Business, through the API and in Antigravity. Researchers are getting a separate carrot: 10,000 scientists will be offered the highest usage tier of Antigravity free for six months, with participants chosen for eligibility and capacity. And Google’s Fairwind Program is expanding to hundreds of additional cyber defense partners, who get the model without the usual guardrails so they can hunt for vulnerabilities.

Pricing is where the launch gets sharp. Argon starts at $2 per million input tokens and $10 per million output tokens, with cached input tokens discounted by 95 percent. That introductory rate only lasts until the end of November. After that, the price doubles to $4 and $20, matching Anthropic’s Claude Opus 5.5 on list price. On a typical task, Artificial Analysis puts Argon at $1.99 during the discount period and $3.98 afterward.

That makes the model look less like a bargain trophy and more like a carefully timed shot across the bow. At the launch discount, Argon lands below GPT-6 Astra on cost while coming in at almost the same performance. Once December arrives, it becomes pricier than GPT-6 Astra and moves ahead of Grok 4.7 in the cost ranking. Still, Google is clearly betting that its pitch will be simple enough: pay now, save now, and hope nobody notices the bill later.

On performance, the company’s own bravado runs into a stiffer independent ranking. Google says Argon leads on benchmarks and tops the Vals Index for finance, coding, law and tax. Artificial Analysis puts it sixth on its Intelligence Index at 52.6 points, just behind GPT-6 Astra at 52.7 and well behind Anthropic’s top entries. The gap to OpenAI is tiny. The gap to Anthropic is not. Google is still trailing the pack it wants to join.