Gemini 4 Argon reaches AI Mode subscribers, and its API price doubles December 1
By Alan Magliocca · Published November 2026
Quick answer: Google made Gemini 4 Argon available on October 13, 2026, in the Gemini API and the Gemini app, and reports say Google AI Pro and Ultra subscribers can also use it in AI Mode. Its introductory API price of $2 input and $10 output per million tokens runs through November 30, 2026, and rises to $4 and $20 starting December 1, so run heavy jobs early and check how AI Mode describes your clients.
Key takeaways
- Gemini 4 Argon became available to developers and to Google AI Pro and Ultra subscribers on October 13, 2026.
- 9to5Google reports that Pro and Ultra subscribers can also use Argon in AI Mode; Google hasn't said free users get it there.
- The API price is $2 input and $10 output per million tokens through November 30, 2026, then $4 and $20 starting December 1.
- Batch and Flex cost half the standard price, and the Gemini API lists no free tier for Argon.
- Check how AI Mode describes each client, and move heavy Argon jobs before December 1.
What happened
On September 30, 2026, Google announced Gemini 4 Argon, its new frontier model for long, complex work in software engineering, enterprise knowledge work such as legal and finance, and cybersecurity defense. Google says it can produce up to 1M tokens of output, up from 64K. It went first to cyber defenders in Google's Fairwind Program.
On October 13, 2026, it opened up more widely:
- Developers: Google said paid-tier Gemini API users can build with Argon in Google AI Studio and Android Studio.
- Gemini app: Google said Google AI Pro and Ultra subscribers can now use Argon in the Gemini app.
- AI Mode: 9to5Google reported that AI Pro and AI Ultra subscribers can also use Argon in AI Mode. Google's Gemini app post doesn't mention AI Mode, and Google hasn't said Argon answers free users' AI Mode searches.
Google's Gemini API pricing page lists Argon's prices per million tokens:
- Standard: $2.00 input and $10.00 output through November 30, 2026. From December 1, 2026: $4.00 input and $20.00 output.
- Batch and Flex: $1.00 input and $5.00 output through November 30, 2026, then $2.00 and $10.00.
- Context caching: $0.10 through November 30, 2026, then $0.20.
- Free tier: none listed.
On benchmarks, Google says Argon scores 77.9% on DeepSWE v1.1, which it calls a new state of the art. That is Google's own reported result.
What it means for agency owners
There are two separate stories here, and they affect different parts of your work.
AI Mode. Paying Google AI subscribers who ask AI Mode about a local service may now get an answer from Argon. Google hasn't said how this changes which businesses get mentioned, so don't promise clients a ranking effect. The groundwork Google has described for AI answers still applies: Google says it has increased inline links in AI Mode and AI Overviews and is testing an "official sites" section. We cover both in Google's AI answers now show more links.
Because Argon is reported for Pro and Ultra users, a client may see different AI Mode answers from a free account and a paid one. Check both views where you can.
The API price. If you run client agents, research jobs or report builders on Argon through the API, the token price doubles starting December 1, 2026. Anything you can run in bulk should run by November 30.
What to do this week
- Check AI Mode for each client. Search their top three service queries, plus their brand name, from a free account and, if you have one, a Pro account. Save screenshots with the date.
- Fix what AI Mode gets wrong at the source: service pages, hours, prices and service area on the client's site and business listings.
- List every Argon API workload you run for clients and estimate December's cost at the new price.
- Move one-off heavy jobs before November 30, 2026. Use Batch for anything that isn't urgent, since it costs half the standard price.
- Test a cheaper model on the same job. Gemini 3.8 Flash may be enough for some of it, and its own introductory price runs through December 31; see Gemini 3.8 Flash's intro price ends December 31.
- Update client quotes that include Argon usage at the December 1 rate.
How AIVX Labs helps
The AI visibility check in AIVX Labs shows whether AI assistants recommend a business, which gives you a baseline before and after you fix a client's pages. The Website Analyzer scores the site out of 100 with fixes in order, and the Fix Package turns those fixes into a deliverable. Tell RICO the job, for example "audit and fix list for a dentist's website," and it picks the tools, shows the credit cost and runs only when you click Go. The website audit and fix package demo shows a full run, and Alan walks through it in this video.
What does Argon cost before and after November 30?
Take an agency running a weekly research-and-report job on Argon for 10 clients. Assume each client's weekly run uses about 400,000 input tokens and 50,000 output tokens, including thinking. Over four weeks, that's 16 million input tokens and 2 million output tokens.
Standard, through November 30, 2026:
- Input: 16 million × $2 = $32
- Output: 2 million × $10 = $20
- Total: $52 a month
Standard, from December 1, 2026:
- Input: 16 million × $4 = $64
- Output: 2 million × $20 = $40
- Total: $104 a month
Batch, from December 1, 2026: $52 a month, the same as the current standard price.
The same token counts on Gemini 3.8 Flash at its introductory standard price come to $12 input plus $7.50 output, or $19.50 a month, rising to $39 from January 1, 2027.
Is Argon still worth it at $4 and $20?
At $4 input and $20 output per million tokens, Argon's standard price from December 1 matches the price Anthropic lists for Claude Opus 5.5. That makes the choice about quality on your job, not price. See Claude Opus 5.5 cut prices about 20% for that side.
A quick way to decide:
- Keep Argon for long, multi-step work where a weaker model makes errors you'd have to fix by hand.
- Move to a cheaper model for summaries, tagging and short drafts, where the output is easy to check.
- Use Batch for anything that can wait a few hours.
- Use caching where you send the same long instructions or documents on every call; cached input is priced at a fraction of fresh input.
- Set an output limit on every call. Argon can produce up to 1M tokens of output, and at $20 per million from December 1, one runaway call that used the full limit would cost $20 on its own.
If you also sell AI Mode visibility to clients who run webinars, BrightStage's guide to Google AI Mode SEO covers that side.
Sources
- Google: Gemini 4 Argon, 2026-09-30
- Google: Build with Gemini 4 Argon, 2026-10-13
- Google: Gemini 4 Argon in the Gemini app, 2026-10-13
- Google AI Studio: Gemini API pricing, 2026-11-11
- 9to5Google: Gemini 4 Argon in the Gemini app, 2026-10-13
Related
Written by Alan Magliocca, founder of AIVX Labs and BrightStage AI (Podavinci LLC). Written with AI assistance and checked against the linked sources.
Read our featured article on LinkedIn
Get AIVX Labs verified for $1: RICO, your AI operator, with 7 days of full access.
Get AIVX Labs Verified for $1Follow along:LinkedInYouTubeTikTokXFacebookInstagramSkool community
Frequently asked questions
Do free AI Mode users get answers from Gemini 4 Argon?
Google's statements and the reports so far refer to Google AI Pro and Ultra subscribers. Google hasn't said free users get Argon in AI Mode.
When exactly does the new API price start?
Google's pricing page lists the introductory price through November 30, 2026, and the new price starting December 1, 2026.
Is there a free tier for Argon in the Gemini API?
No. Google's pricing page lists no free tier for Gemini 4 Argon, and Google says developer access is for paid-tier API users.
