AI-rewritten: This is a summary of an article from Ars Technica, rewritten by AI (Qwen, running locally) to make it easier to read. The facts come from the original article – read it for the full story.

---

Google is rolling out Gemini 4 Argon today in the Gemini app, but only for paying users. The company states it is moving cautiously due to concerns about potential misuse, yet access has already begun for those on AI Pro ($20/month) and AI Ultra (starting at $99/month) plans. Lower-tier subscribers will not have full access at this time.

The model retains its previously announced capabilities in reasoning and coding. Google claims Argon is trained to refuse malicious instructions and includes an automated monitor that stops processing if it detects unintended workflows. Inside Google, the AI has reportedly improved YouTube’s comment system efficiency by eliminating redundant emoji rendering, which dropped server load by 45 percent. It also found critical bugs in a drone physics simulator used in Project Wing that had existed since 2018.

Access to Argon is restricted further for coding-focused products like Antigravity and AI Studio, where only AI Ultra subscribers can use it. While Pro users can generate code in the Gemini app’s Canvas mode, more complex multi-threaded workflows in Antigravity are currently too expensive. For developers using the paid API, pricing is temporarily $2 per million input tokens and $10 per million output tokens, with a 95 percent discount for cached tokens. This offer ends on November 30, after which rates will double to $4 and $20 respectively.