Google Ships Gemini 4 Argon Worldwide, Leaves 3.5 Pro in the Dust
Google flipped the switch on Gemini 4 Argon globally on October 13, bringing its new top-tier model to paid users in Japan. The model it was meant to replace never made it out the door — Google told Reuters in September it has no plans to release Gemini 3.5 Pro. Developers have until November 30 before API prices double.
- Original source: note.com.
- The source published it on Nov 14, 2026 at 11:38 PM.
- No second outlet has reported it yet. Treat it as a single-source claim.
Google's new flagship is out the door — just not for everyone yet. The company began rolling out Gemini 4 Argon globally on October 13, 2026. In Japan, it shows up inside the Gemini app for people on the paid Google AI Pro and Google AI Ultra plans. Developers and enterprises get it through the paid Gemini API tier, Google AI Studio and the Gemini platform, while Ultra subscribers also get access through Google's development tool Antigravity. Pro and Ultra users can reach it through the Gemini app and AI mode in Google Search.
Argon arrived first on September 30 with a deliberately narrow door: access went only to vetted cybersecurity personnel through a framework called the Fairwind Program, and Google said it would take part in the US government's voluntary pre-access framework. A Google spokesperson told Reuters that day that Argon is larger in scale than previous Pro models. Google says Argon's output limit is now 1 million tokens, up from a previous ceiling of 64,000. One internal example: migrating huge C and C++ programs to Rust, including targets exceeding 800,000 lines.
The pricing clock is ticking. Developers pay per million tokens: input currently runs $2, output $10, with a 95% discount on the input rate when the same input is cached. Those introductory rates hold until November 30 — then input goes for $4 and output for $20. As of November 15, roughly two weeks remained before the hike. App users aren't billed per token; Argon sits inside monthly plans, with Google AI Pro at 2,900 yen a month and Google AI Ultra at 14,500 yen or 32,000 yen a month.
Here's the twist: the model Argon effectively replaced never shipped. Google announced Gemini 3.5 Pro at its I/O developer conference on May 19, 2026, saying it would arrive the following month. June came and went. Bloomberg reported on July 16 that the model was months behind internal schedule, citing improvements in coding as one reason. On September 30, a Google spokesperson told Reuters there were no longer any plans to release Gemini 3.5 Pro — a shelving Google never mentioned on its official blog. Google's own explanation for the delays has gone no further than "to improve performance."
While the Pro line stalled, Google flooded the zone with speed-focused models. It pushed out Gemini 3.6 Flash, Gemini 3.5 Flash-Lite and Gemini 3.5 Flash Cyber for cybersecurity on July 21 — the same day it said 3.5 Pro was being tested with partners and revealed pre-training had begun on Gemini 4. Gemini 3.7 Flash landed August 13, and Gemini 3.8 Flash followed September 2, the third Flash release in six weeks, with Google describing it as better performance at the same price as 3.7 Flash.
Argon is the only model carrying the Gemini 4 name, and Google hasn't explicitly called it the successor to Pro. The Gemini app's model picker now lists 3.5 Flash-Lite, 3.8 Flash, 3.1 Pro and 4 Argon — the Pro slot empty, with Argon sitting above it. The staged rollout was intentional: Google says it will shore up safety in four areas before widening access, including preventing misuse tied to cyberattacks or chemical, biological, radiological and nuclear materials, guarding against indirect prompt injection, monitoring the AI's chain of thought and halting it when it drifts from user intent, and isolating the work environment before dangerous training or evaluation.
TopicsGoogle · Gemini 4 Argon · Gemini 3.5 Pro · Reuters · Bloomberg · Gemini 3.8 Flash · Google Antigravity · Google AI Ultra
Source: note.com
Written by AI from verified reporting. Not financial advice.

