Two dollars in, ten dollars out. That's what Gemini 4 Argon, which Google opened to everyone on October 13, charges today in the API per million tokens. Those rates hold until November 30. On December 1, both double to $4 and $20, which means anyone who builds an automation on this model today will see twice the bill for the same work five weeks from now.
The math on the app side is just as clear. In Turkey, the cheapest way to use Argon in the Gemini app is Google AI Pro at 869.99 Turkish lira (TRY) a month. It's not on free accounts, and it's not on the API's free tier either. Frankly, this time Google chose to keep its best model behind a door. So the natural question is: is it worth walking through?
What changed?
Google first announced the model on September 30, but opened it only to cybersecurity teams in a program called Fairwind [A]. General availability came two weeks later, on October 13: the same day saw both the Gemini app announcement and the "generally available" entry in the API changelog [A]. The Gemini API changelog describes Argon as "our most capable model for complex coding, agentic, math, and science tasks" [A]. The model ID is gemini-4-argon; in the API, input and output share a single 1 million token limit [A].
What changed in the Gemini app goes beyond the model. According to Google's announcement, Argon answers more directly and tells you what it's doing at intervals during long jobs. There's a button that switches from chat to a "tasks" mode: Gemini can keep working across Google apps while you're away from your computer. The number of apps it can connect to has passed 40 (Instacart, OpenTable, Adobe, and Shopify are given as examples) [A]. Which of these connections work in Turkey is not clear from the official source; it's not hard to guess that Instacart has no equivalent there.
Then there's the coding side. Google Antigravity(in Turkish) now lives inside Gemini's macOS app [A]. At first, Argon in Antigravity was open only to Ultra subscribers, and for two days Google's developer forum filled up with complaints from Pro subscribers. The official reply from Google on October 16: Pro subscribers can also use Argon in Antigravity, but only with a separate quota that resets weekly. The same reply also says Gemini 3.1 Pro, 3.6 Flash, and 3.7 Flash have been removed from Antigravity [A].
There's no news for Gemini CLI(in Turkish) users. The GitHub issue requesting Argon support was opened on October 9, and as of October 28 it's still open with no reply from the maintainers [C].
Source: Gemini API pricing page, as of October 28, 2026.
Who can use it, where, and at what price?
In Turkey: the Gemini subscription page looks like this as of October 28, 2026 [A]:
Plan
Monthly price (Turkey)
Gemini 4 Argon
Usage limit
Free
₺0
No, Flash-Lite
Standard
Google AI Plus
₺199.99
Page says Flash
2x standard
Google AI Pro
₺869.99
Yes
4x standard
Google AI Ultra
From ₺1,479.99 (top tier ₺8,999.99)
Higher access
5x or 20x Pro
Limits reset every 5 hours, with a weekly cap on top; on a subscription, once you hit the limit the chat continues with Flash-Lite [A].
There's a snag here. On Google One's general plans page, the AI Plus card says "Includes Gemini 4 Argon," while the Turkey subscription page lists only Flash for Plus. The October 13 announcement doesn't mention Plus at all [A]. Which one is current is not clear from the official sources. If you're thinking of switching to Plus just for Argon, check the model picker in your own account first.
The API side is simpler:
Usage
Until November 30, 2026
From December 1, 2026
Standard input / output
$2 / $10
$4 / $20
Batch and Flex input / output
$1 / $5
$2 / $10
Cached input (automatic caching only)
$0.10
$0.20
Grounding with Google Search
5,000 requests a month free, then $14 per 1,000 requests
Same
Free tier
None
None
US dollars per million tokens; output includes thinking tokens. Source: Gemini API pricing page [A].
There's one more line on the pricing page you shouldn't miss: on the paid standard tier, the "used to improve our products" column says "No." On the cheaper, invitation-only Contributor tier, it says "Yes" [A]. If you'll be sending customer data, that's the price of the cheap option.
Who is it for?
If you have a developer or write code yourself: Argon inside Antigravity is now something you can try with a Pro subscription. You won't know what happens when the weekly quota runs out until you've used it for a week.
Offices that read long documents, contracts, and financial spreadsheets: these are the areas Google highlights, and in the independent Vals benchmark Argon scores high on legal and finance [C]. How it performs on Turkish-language documents, however, is not clear from the official source.
Agencies and small businesses building automations with the API: since the price changes on December 1, budget with $4/$20, not today's rates.
People getting by with free Gemini: you can skip it. For everyday messages, summaries, and email drafts, I think you mostly won't notice Argon's difference. This model earns its keep on long, multi-step work.
Gemini subscription (Pro, ₺869.99)
Chat, tasks mode, and Antigravity in one package
Fixed bill, no counting tokens
Limit resets every 5 hours, with a weekly cap
When you hit it, you drop to Flash-Lite
Gemini API (paid tier)
Connects to your own automation or app
Pay as you go; the price doubles on December 1
Batch is half price for bulk jobs
No free tier; you have to enable billing
Which door should you use Argon through?
Where does it stand next to ChatGPT and Claude?
Google's October 13 announcement described Argon as number one on the Vals Index [A]. In Vals's ranking updated October 27, it's now third: two Anthropic models have moved ahead of it [C].
Benchmark
Gemini 4 Argon
Claude Sonnet 5.5
Claude Opus 5.5
GPT-6 Astra
GPT-6.1 Sol
Vals Index (independent, October 27)
69.24%
70.85%
70.18%
65.02%
63.94%
Terminal-Bench 4.0 (Google's table)
57.4%
n/a
66.4%
58.2%
n/a
DeepSWE v1.1 (Google's table)
77.9%
n/a
74.2%
74.1%
n/a
Vals Index: vals.ai, October 27, 2026 [C]. The other two rows come from the table in Google's model card, part of which Google calculated itself; Sonnet 5.5 and GPT-6.1 Sol are not in that table [C].
Here's my read: on the Vals Index, the top models from the three companies are within a point or two of each other, and the order changes almost every week. "Which model is best" doesn't tell you much this month; you need to ask which job you're doing. For a team that has the model write code in the terminal, Opus 5.5's nine-point lead on Terminal-Bench is serious; for someone working with documents hundreds of pages long, Argon's 1 million token window may matter more. Which one weighs more in your work? If you're curious what changed on the free tier, read GPT-6 in ChatGPT and free users; if you're looking for a cheap model, read Claude Haiku 5.5.
Before you pay for Argon
0/5
Check the items; the empty ones are where to look before you decide.
People generating images; details at the link below
We covered Nano Banana 2.1's text-in-image and character consistency side in a separate post.
FAQ
Is Gemini 4 Argon available on a free Gemini account?
No. According to the Turkey subscription page, the free plan runs on Flash-Lite, and Argon is on Pro and Ultra. Google hasn't said when it will open it to free users; I'm not going to speculate.
Can you use Gemini 4 Argon in Turkey, and does it know Turkish?
The subscription page in Turkey lists Argon on the ₺869.99 Pro plan, and the October 13 release notes were published in Turkish, so access with a subscription appears to be open. A separate measurement of its Turkish-language performance is not clear from the official source. Which connected apps work in Turkey isn't stated either.
Can I try it for free in the API?
No. The pricing page states clearly that Argon is not on the free tier; you can't make calls without attaching billing to the project. Batch gets you half price on bulk jobs, but that's for work that isn't urgent.
If you want to talk through, using your own business, which jobs make sense on a subscription and which on the API, see the AI consulting page; if you want to automate a repetitive task with the Gemini API, see the AI automation page. To see where you stand first: START YOUR FREE ANALYSIS .
Beyond GPT-6, ChatGPT got five new features in October 2026. Audio uploads are on paid plans, Slides on Pro and business plans, locked chats for everyone; Finances is US-only.
OpenAI released GPT-6.1 Terra on October 26, 2026: on the API, one million input tokens cost $1 and output tokens $5, half the price of its predecessor. It's available in ChatGPT Work and Codex on Plus and higher plans.
Google has deprecated Gemini 2.5 Pro and Flash; starting December 1, Gemini API requests are quietly redirected to 3.8 Flash and 3.5 Flash-Lite. What happens to price and quality, and which automations are affected?
+2
We use cookies to improve the site and measure advertising performance. Analytics and advertising cookies run only if you accept them. Details are in our Privacy Policy and Data Protection Notice.