Google made Gemini 4 Argon generally available on October 13, with the model, gemini-4-argon, reachable through the API, AI Studio, Antigravity on the Ultra tier, and the Gemini app under Pro and Ultra. Intro pricing runs $2 per million input tokens and $10 per million output tokens through November 30, 2026. After that, the rate doubles to $4 and $20. That is the whole announcement. Google's post does not say why the window is six weeks, or what triggers the jump beyond the calendar date itself.
The model carries a 1 million token unified context window, long decode, selectable thinking levels, multimodal input, code execution, and function calling. None of those features are new concepts industry-wide, but the bundle matters for teams that already route traffic by context size or cost per call. A service built around Gemini's prior pricing tier will see token costs double on a fixed date, not gradually, which changes how finance and engineering should model spend for Q4 and Q1 budgets.
Google has confirmed availability across three surfaces and the two-tier pricing schedule. It has not disclosed rate limits, regional rollout order, deprecation plans for prior Gemini versions, or whether Antigravity access on Ultra differs functionally from API access. Teams relying on fallback chains that include an older Gemini model should check whether that model remains supported past this release, since GA announcements sometimes precede quiet sunset notices for predecessors. None of that is confirmed here either way.
For platform operators, the immediate task is inventory, not adoption. Check which services call Gemini models by version string versus an alias that could silently resolve to Argon. Check whether cost alerts are keyed to per-token rates that will double on November 30, because a budget that holds at $2/$10 breaks quietly at $4/$20 if nobody updates the threshold. Check whether retry and timeout logic assumes the long-decode behavior of a prior model, since function calling and code execution paths often have different latency profiles under load, and that mismatch shows up first in production, not in testing.
The six-week window is short enough that any migration decision needs to happen this month, not next quarter. Teams that delay past November 30 absorb a 100 percent per-token price increase automatically, with no indication from Google that the deadline will slip.