English · 6 min read
Gemini 4 Argon pricing after the promo: what it costs from 1 December 2026
Gemini 4 Argon is generally available at $2 / $10 per 1M tokens, but only until 30 November 2026. From 1 December it doubles to $4 / $20. Here is what that does to a small business bill, and how it compares with Claude and GPT.
By Dapols ·
The short answer: Gemini 4 Argon costs $2 per 1M input tokens and $10 per 1M output tokens through 30 November 2026, then $4 and $20 from 1 December 2026. That makes it the same price as Claude Sonnet 5.5 and GPT-6.1 Sol today, and the same price as Claude Opus 5.5 from December. If you build anything on it, budget for the December price, not the launch price.
Prices are from Google's Gemini API pricing page, Anthropic's pricing page and OpenAI's API pricing page, checked on 20 October 2026. Totals below are our own arithmetic from those list prices, not measured bills.
What changed on 13 October
Google announced Gemini 4 Argon on 30 September 2026, first only for vetted cyber defenders. Its Gemini API release notes list gemini-4-argon as generally available from 13 October 2026 (Gemini API changelog). Google's launch post describes the $2 / $10 rate as an introductory price, with cached input 95% off, and says it rises to $4 / $20 afterwards (Google). The pricing page now gives the dates: the introductory rate runs through 30 November, and the higher rate starts 1 December.
The price on both sides of the date
| Input per 1M tokens | Output per 1M tokens | |
|---|---|---|
| Gemini 4 Argon, through 30 Nov 2026 | $2 | $10 |
| Gemini 4 Argon, from 1 Dec 2026 | $4 | $20 |
| Claude Sonnet 5.5 | $2 | $10 |
| GPT-6.1 Sol | $2 | $10 |
| Claude Opus 5.5 | $4 | $20 |
So the promotion makes Argon look like a mid-price model. From December it is priced like Anthropic's top model.
What that means in dollars
Take a small business that sends 5 million tokens in and gets 1 million tokens out each month (roughly a few hundred drafted documents and summaries). At list price:
- Gemini 4 Argon in November: 5 x $2 + 1 x $10 = $20
- Gemini 4 Argon from December: 5 x $4 + 1 x $20 = $40
- Claude Sonnet 5.5 or GPT-6.1 Sol: $20, before and after
- Claude Opus 5.5: $40
The jump is a doubling. For a workload this small the dollar difference is modest, but the same ratio applies if you scale it 100 times.
Is it worth the December price?
On the one independent score we have, not by itself. Artificial Analysis scored Gemini 4 Argon 53 on its Intelligence Index in our 1 October capture, against 58 for Claude Opus 5.5 and 56 for Claude Sonnet 5.5, with GPT-6.1 Sol at 52. Google's own headline number is 77.9% on the DeepSWE coding benchmark, which Google calls a new state of the art. That is Google's figure, not an independent test.
Scores are not bills, either: models differ in how much they write, and output is where the money goes. Artificial Analysis measured about $0.72 per index task for GPT-6.1 Sol and about $7.60 for Claude Sonnet 5.5 at max effort, which shows how far "same price per token" is from "same price per job". We have no equivalent cost-per-task figure for Argon yet.
What to do
- Test it during the promo. If you have a hard coding or analysis task, run it on Argon this month and record the real token counts.
- Compare at the December price. Price your test at $4 / $20 and ask whether it beats Claude Sonnet 5.5 or GPT-6.1 Sol at $2 / $10 on the same job.
- Do not hard-code the promo rate into a quote, a client estimate or a budget that runs past 30 November.
- Use the cheap tier for volume. If most of your calls are routine drafting or sorting, a low-cost model such as Claude Haiku 5.5 or GPT-6 Luna is a better home than any flagship.
You can see Argon next to its rivals on our Gemini 4 Argon vs Claude Opus 5.5 and Gemini 4 Argon vs GPT-6.1 Sol pages, and every recent release is on the AI model tracker.
Sources: Gemini API pricing, Gemini API changelog, Google: Gemini 4 Argon, Anthropic pricing, OpenAI API pricing, Artificial Analysis: Gemini 4 Argon