The Revenue AI Report
← Reality Check

Gemini 4 Argon Is Live: Price Doubles Dec 1. Write the Order Form at $4/$20.

Google made Gemini 4 Argon generally available Oct 13, 2026 at an intro $2/$10 per 1M tokens through Nov 30. From Dec 1 it is $4/$20. Five order-form checks.

Jonathan Kvarfordt · Published October 13, 2026 · Last updated October 13, 2026 · 8 min read

Why trust this analysis?

The short answer

What's in it for you: both Argon prices, the date the higher one starts, the meters that bill on top of the rate card, and five lines to write into any order form, pilot plan or business case that runs on Argon.

Budget Gemini 4 Argon at $4 per million input tokens and $20 per million output tokens, the rate Google lists from Dec 1, 2026. Treat $2/$10 as a seven-week launch rate. Output includes thinking tokens, cached input doubles on the same day, and search grounding bills per query. Put the Dec 1 rate, a vendor pass-through answer and a cost cap in the order form.

Decision rule

Approve no Argon business case that shows only the launch rate.

Operator action

This week: find every quote, pilot plan and vendor proposal that names Argon. Re-price each one at $4/$20 and send the vendor the pass-through question below before you sign.

Last reviewed

Google put Gemini 4 Argon on sale on Oct 13. The sale ends Nov 30.

Google made Gemini 4 Argon generally available on Oct 13, 2026. Its introductory API price is $2 per million input tokens and $10 per million output tokens through Nov 30, 2026 (Google launch; Gemini API pricing). A footnote on the launch post sets the end date: "After the introductory period expires on November 30, 2026, the price of $4 per 1M input tokens and $20 per 1M output tokens will apply." (Google launch) Any pilot, vendor quote or internal business case priced on Argon this month is priced on a rate with about seven weeks left.

What did Google announce on Oct 13?

Google moved Gemini 4 Argon from limited access to general availability and put a hard end date on the launch price. The launch post, timestamped 10:00 AM MT, lists Argon in the Gemini API, AI Studio, Gemini for business and Google Antigravity, and in the Gemini app for Google AI Ultra and Pro subscribers (Google launch). For companies, the post says: "Try it in Gemini for business, the API, and in Google Antigravity." (Google launch)

Two things changed since Sept 30. That day Google announced Argon with access limited to its Fairwind Program and the same $2/$10 introductory price, with no end date (Google Sept 30). On Oct 13, Argon opened to buyers, and the launch price now expires Nov 30, 2026 (Google launch). VentureBeat covered the general release at 10:08 AM MT (VentureBeat).

Check your plan before you scope a pilot. Google AI Pro and Ultra subscribers get Argon in the Gemini app and AI Mode in Search. Inside AI Studio and Antigravity, subscriber access is Ultra only. Developers and companies can use it on the paid Gemini API tier, which is billed separately from any subscription; there is no free API tier for Argon (Google launch; Gemini API pricing; Google AI plans). Price any pilot at API rates, even if the team first tried Argon in the Gemini app.

What does Gemini 4 Argon cost before and after Dec 1?

Input, output and cached input all double on Dec 1, 2026 (Gemini API pricing).

Meter, per 1M tokens (Standard, paid tier)Through Nov 30, 2026From Dec 1, 2026
Input$2.00$4.00
Output, "including thinking tokens"$10.00$20.00
Cached input (context caching)$0.10$0.20

Source: Gemini API pricing page, Standard tab, last updated Oct 13, 2026 UTC (Gemini API pricing). Cached input runs 95% off the input rate before and after Dec 1 (Google launch; Gemini API pricing).

Google also lists Batch, Flex and Priority rates for Argon, each with its own Dec 1 step. This decoder uses the Standard rates (Gemini API pricing).

Step chart of Gemini 4 Argon Standard API prices per million tokens. From Oct 13 to Nov 30, 2026, input is $2 and output is $10. From Dec 1, 2026, input is $4 and output is $20. Output includes thinking tokens.
Google's launch rate for Gemini 4 Argon ends Nov 30. Budget at the Dec 1 rate. Source: Google, Oct 13, 2026.

Which meters move the bill past the rate card?

Three meters decide what an Argon task costs on top of the input rate.

Thinking tokens. Google's pricing page labels the output row "Output price (including thinking tokens)." (Gemini API pricing) When the model reasons through a hard prompt, that reasoning bills at the output rate: $10 per million through Nov 30 and $20 from Dec 1. Two prompts of the same length can cost very different amounts, and the gap shows up only in your usage logs.

Caching. The pricing page lists "Implicit caching only (explicit caching is not supported)." (Gemini API pricing) Our reading: Google applies the cached rate when it detects repeated input, and you cannot reserve it for a long system prompt or a price sheet. Model your business case on a cache hit rate you have measured. Until you have one, assume zero.

Search grounding. Grounded answers draw on Google Search. On the Standard rate table, grounding includes 5,000 free requests a month, shared across Gemini 3 and newer models, then costs $14 per 1,000 (Gemini API pricing). Google says one request "may result in one or more queries to Google Search" and that it charges "for each individual search query performed." (Gemini API pricing) One question from a seller, an underwriter or a service rep can fire several billable searches. Any other Gemini work on the same account draws from the same free pool.

Why does a launch price break a revenue business case?

A pilot that proved ROI at $2/$10 proved it at a rate that ends Nov 30. The receipt has to hold at the Dec 1 meter.

For most teams, the people who check AI output cost more than the tokens. We made that case in AI review cost. A doubled model rate still breaks a business case that was built on the launch number, and the damage lands in three places.

  • Your own build. An insurer's quote-intake agent, a distributor's account research agent or a bank's call-summary tool priced in October carries a cost step on Dec 1. For a calendar fiscal year, that step lands inside Q4.
  • Vendor tools. Research agents, content agents and per-seat AI tools may run on frontier models. If a vendor priced its seats on Argon's launch rate, its Argon cost at list rates doubles on Dec 1. Someone pays for that. Find out who before you sign.
  • Model choice. At list API rates, Claude Sonnet 5.5 and GPT-6.1 Sol are $2/$10 per million tokens, and Claude Opus 5.5 and Meta Muse Wonder are $4/$20. OpenAI's figures are its short-context rates (Anthropic pricing; OpenAI pricing; Meta Muse). Argon matches the first pair through Nov 30 and the second pair from Dec 1. A team that picks Argon as its default this month should price that default with the $4/$20 group.

Five order-form checks for Gemini 4 Argon

  1. Price it at the Dec 1 rate. Show two lines in the business case: the launch rate through Nov 30 and $4/$20 from Dec 1. Approve on the Dec 1 line. The board line reads: "Our AI costs are modeled at the Dec 1 list rate."
  2. Ask the pass-through question. For any vendor that runs on Argon, ask in writing: "Does your per-seat or per-credit price assume Google's introductory rate? Is our price locked for the term when your model cost doubles on Dec 1?" Get the notice period for any price change into the contract.
  3. Cap the thinking. Output bills with thinking tokens included. Ask the vendor, or your own team, for a cost-per-task estimate at the Dec 1 output rate. Set an output cap per task in code and log every task that hits it.
  4. Put a ceiling on grounding. Search grounding bills per query, and one request can fire several. Cap queries per request and per month, alert when the free 5,000 runs out, and decide who approves spend past the cap.
  5. Re-run cost per accepted result before the production commit. An accepted result is output a person kept: an approved quote, a brief a seller used, a summary nobody rewrote. Re-price the pilot's accepted results at Dec 1 rates, add review time, and sign the production commit on that number.
Checklist of five order-form lines before signing for an AI tool that runs on Gemini 4 Argon: price it at the Dec 1 rate, ask the vendor the pass-through question, cap thinking tokens per task, put a ceiling on search grounding queries, and re-run cost per accepted result before the production commit.
Five lines to write before any pilot or vendor quote on Gemini 4 Argon goes to production.

Who owns which line?

  • CRO, CEO or founder: sign only on Dec 1 math, and take the board line above into the next review.
  • Revenue finance: book the Dec 1 cost step in the forecast now. On a calendar fiscal year it lands in Q4.
  • RevOps and GTM engineering: model choice, routing, the output cap, the grounding ceiling and the usage logs.
  • Sales and marketing leaders: ask the pass-through question at every renewal for research and content tools that run on frontier models.
  • Enablement, customer success and partnerships: per-seat AI tools carry the vendor's model costs. Add one pass-through question to your next renewal.

One more dated step: Gemini 3.8 Flash

Gemini 3.8 Flash has a dated step too. Its paid Standard rate is $0.75 in and $3.75 out per million tokens through Dec 31, 2026, then $1.50 and $7.50 from Jan 1, 2027 (Gemini API pricing). Check every Gemini model in your stack for an end date.

Soft opens

  • Vendor claims. Google says Argon ranks #1 on the Vals Index, and that "Argon in Antigravity has helped teams increase agentic code submissions by more than 35% while reducing code rollbacks by nearly a third" (vendor-reported) (Google launch). Vals AI's own leaderboard shows Argon first at 68.90% as of Oct 12, 2026, ahead of Claude Sonnet 5.5 at 67.04% and Claude Opus 5.5 at 66.97% (Vals Index). The code figures come from Google's own coding work. Test Argon on your own revenue workflow before you count on them.
  • Grounding unit. The pricing row prices grounding "per 1,000 requests," and its footnote bills "each individual search query performed." (Gemini API pricing) Ask Google how one request with several searches counts against the free 5,000 and the $14 rate.
  • List prices only. Committed-spend or enterprise agreements may price differently. Google can change its pricing page. Check it on the day you sign.

Related decoders

Sources

  1. Google, "Gemini 4 Argon: Built to solve your hardest problems," Koray Kavukcuoglu, SVP, Google DeepMind, Oct 13, 2026, 10:00 AM MT: blog.google
  2. Google, Gemini API pricing (Gemini 4 Argon and Gemini 3.8 Flash, Standard tab, paid tier), last updated Oct 13, 2026 UTC: ai.google.dev
  3. VentureBeat, "Google makes Gemini 4 Argon broadly available with limited-time low API pricing of $2/$10 per 1M token," Oct 13, 2026, 10:08 AM MT: venturebeat.com
  4. Google, Gemini 4 Argon announcement, Sept 30, 2026: blog.google
  5. Google, Google AI plans for the Gemini API, last updated Oct 13, 2026 UTC: ai.google.dev
  6. Anthropic, Claude pricing, "Model pricing" table, Oct 2026: platform.claude.com
  7. OpenAI, API pricing, Flagship models, Standard tab, short context, Oct 2026: developers.openai.com
  8. Meta, "Put your ideas to work with Muse Wonder and Muse Wonder Full," Josh Walters, Oct 12, 2026: dev.meta.ai
  9. Vals AI, Vals Index, updated Oct 12, 2026: vals.ai

New order-form decoders land in the weekly issue first. Subscribe.

Frequently asked questions

How much does Gemini 4 Argon cost?
Google lists an introductory $2 per million input tokens and $10 per million output tokens through Nov 30, 2026. From Dec 1, 2026, it lists $4 and $20. Output includes thinking tokens.
When does the Gemini 4 Argon introductory price end?
Google's launch post says the introductory period expires on Nov 30, 2026, and the $4/$20 price applies after that.
Can Google AI Pro subscribers use Gemini 4 Argon?
Yes, in the Gemini app and AI Mode in Search. Inside AI Studio and Antigravity, subscriber access is Ultra only. API use runs on the paid Gemini API tier and is billed separately from any subscription.
Does cached input get cheaper on Gemini 4 Argon?
Google lists context caching at $0.10 per million tokens through Nov 30 and $0.20 from Dec 1, 95% off the input rate, with implicit caching only. Model costs on a cache hit rate you have measured.
How is Google Search grounding billed with Gemini 4 Argon?
The Standard rate table lists 5,000 free requests a month, shared across Gemini 3 and newer models, then $14 per 1,000. Google charges for each search query, and one request can produce several queries.
What should a revenue team put in an Argon order form?
The Dec 1 rate as the approval number, a vendor pass-through answer with a price lock, an output cap per task, a ceiling on grounding queries, and cost per accepted result re-run at Dec 1 prices.

Share this page

Posting to Instagram or TikTok? Copy the link, it carries the title, summary and share image.

Keep reading