Disclosure: AI Tools Vetted is reader-supported and may show ads on this page (including Adcash). Google has no affiliate program with us, so every Google link here is a plain link and earns us nothing. Sponsored add-on links, if any, sit in a separate box at the end and are not part of this article’s analysis.
Last updated October 16, 2026 · Prices checked October 16, 2026 on Google’s official Gemini API pricing page (USD).
Gemini 4 Argon is now generally available
TL;DR
- GA: October 13, 2026 — globally available for developers and enterprises via the API in AI Studio and the Gemini platform; Google AI Ultra subscribers in Google Antigravity; Google AI Ultra and Pro subscribers in the Gemini app and AI Mode in Search (Google blog, Oct 13, 2026).
- Model ID:
gemini-4-argon(Stable) on the Gemini API models page. - Intro API price (through Nov 30, 2026): $2 per 1M input / $10 per 1M output (Standard); doubles to $4 / $20 on December 1, 2026.
- API Free tier: not available. Pricing page: “not available on the Free tier or on the Google AI Pro plan.”
Google announced Gemini 4 Argon on September 30, 2026 with restricted access for trusted cyber defenders through the Fairwind Program (original announcement). On October 13, 2026 it moved to general availability for the audiences listed above (GA post). This article covers status and cost as of October 16, 2026 — not the September launch as if it were new today.
Who can use it and where
Per Google’s October 13 blog:
| Audience | Where Argon is available (per Google blog) |
|---|---|
| Developers & enterprises | Gemini API in AI Studio and Gemini platform |
| Google AI Ultra subscribers | Google Antigravity; also Gemini app and AI Mode in Search |
| Google AI Ultra and Pro subscribers | Gemini app and AI Mode in Search |
Source: Introducing Gemini 4 Argon (GA), checked October 16, 2026.
The Gemini API pricing page adds an important limit for API access: Gemini 4 Argon is not available on the Free tier or on the Google AI Pro plan (API pricing). Reading both sources together, without going beyond them: Google AI Pro subscribers get Argon in the Gemini app and AI Mode in Search (blog), while API access is billed on the paid API tier and is not offered on Free or on the Google AI Pro plan for API (pricing page). We are not asserting anything about AI Studio UI labels beyond that.
Consumer Google One / AI subscription dollar amounts are omitted here — we did not open Google’s consumer plan page on October 16 for this article.
Pricing: introductory rate until November 30, 2026
Google’s GA blog states an introductory price of $2 per million input tokens and $10 per million output tokens, with cached input tokens at 95% off, and a footnote that after November 30, 2026 the price becomes $4 / $20 per 1M input / output (GA blog footnote).
The Gemini API pricing page for gemini-4-argon (paid tier, USD per 1M tokens) matches that schedule for Standard, and lists Batch and Flex at half the Standard rates (pricing, last updated 2026-10-15 UTC per page footer):
| Rate (USD / 1M tokens) | Through Nov 30, 2026 | From Dec 1, 2026 |
|---|---|---|
| Standard input | $2.00 | $4.00 |
| Standard output (incl. thinking) | $10.00 | $20.00 |
| Standard context caching (implicit only) | $0.10 | $0.20 |
| Batch input | $1.00 | $2.00 |
| Batch output (incl. thinking) | $5.00 | $10.00 |
| Batch context caching | $0.05 | $0.10 |
| Flex input | $1.00 | $2.00 |
| Flex output (incl. thinking) | $5.00 | $10.00 |
| Flex context caching | $0.05 | $0.10 |
Grounding with Google Search or Maps: 5,000 free requests per month (shared across Gemini 3 and newer models), then $14 per 1,000 requests (pricing).
Free tier for Argon: Not available.
Specs Google publishes
| Spec | Gemini 4 Argon (official) |
|---|---|
| Model ID | gemini-4-argon (Stable) |
| Context / decode | Unified 1M context window; long decodes up to 1M tokens (models); blog: “1M unified input-output token limit via the Gemini API” |
| Positioning | Flagship reasoning model for complex software engineering, deep math/science, autonomous multi-step agents (pricing) |
What Google says it’s good at
These are Google’s claims from the October 13 GA post, not our tests:
- Google reports Argon ranks #1 on the Vals Index (finance, coding, legal, tax work weighted by economic sector).
- Google reports internal use cases in large codebase refactoring, cybersecurity defense (Fairwind partners), quantum/fusion/rare-disease research, and robotics (87% on Embodied Reasoning / ERQA “relying on pure reasoning,” per Google).
- Google reports agentic code submissions in Antigravity up more than 35% with code rollbacks down nearly a third (internal Google engineering).
We have not run independent benchmarks. Do not treat vendor scores as cross-vendor comparisons.
API price snapshot vs Claude Opus 5.5 and GPT-6 (status only)
No benchmark mixing. Official Standard short-context API list prices only, checked October 16, 2026:
| Model (API) | Input / 1M | Output / 1M | Notes |
|---|---|---|---|
Gemini 4 Argon (gemini-4-argon) |
$2.00 intro → $4.00 from Dec 1 | $10.00 → $20.00 | Not on Free API tier (Google pricing) |
| Claude Opus 5.5 | $4.00 | $20.00 | (Anthropic pricing) |
GPT-6.1 Sol (gpt-6.1-sol) |
$2.00 ($0.10 cached) | $10.00 | (OpenAI API pricing; also GPT-6.1 Sol post) |
GPT-6 Sol (gpt-6-sol) |
$2.00 ($0.20 cached) | $10.00 | (OpenAI API pricing) |
Availability and product surfaces differ (ChatGPT vs Gemini app vs Claude). Price rows are not a recommendation.
What this changes for you
The Dec 1 doubling is the decision that matters for budgets. The math below is ours, using only the official Standard / Batch / Flex rates above. Thinking tokens count as output on Google’s page — if your jobs emit large thinking blocks, real bills run higher than these text-only sketches.
Example 1: 1M input + 200K output (one large job)
| Path | Through Nov 30 | From Dec 1 |
|---|---|---|
| Standard | (1 × $2) + (0.2 × $10) = $4.00 | (1 × $4) + (0.2 × $20) = $8.00 |
| Batch or Flex | (1 × $1) + (0.2 × $5) = $2.00 | (1 × $2) + (0.2 × $10) = $4.00 |
Same work costs twice as much after Dec 1 on every row. Batch/Flex stays at half of Standard either side of the cliff.
Example 2: 500K input + 50K output (mid-size request)
| Path | Through Nov 30 | From Dec 1 |
|---|---|---|
| Standard | (0.5 × $2) + (0.05 × $10) = $1.50 | (0.5 × $4) + (0.05 × $20) = $3.00 |
| Batch or Flex | (0.5 × $1) + (0.05 × $5) = $0.75 | (0.5 × $2) + (0.05 × $10) = $1.50 |
Example 3: monthly volume — 10M input + 2M output
| Path | Through Nov 30 | From Dec 1 |
|---|---|---|
| Standard | (10 × $2) + (2 × $10) = $40 | (10 × $4) + (2 × $20) = $80 |
| Batch or Flex | (10 × $1) + (2 × $5) = $20 | (10 × $2) + (2 × $10) = $40 |
Practical takeaways
- Budget now at $4 / $20 Standard if the workload continues past November 30. Intro rates are temporary.
- Move non-urgent jobs to Batch or Flex before and after Dec 1 — same 50% cut relative to Standard on the published tables.
- Who should wait: if you only need Argon in the Gemini app as a Pro/Ultra subscriber, API pricing may not affect you; we are not quoting consumer subscription prices here.
- Who should act this month: paid API teams with predictable bulk jobs. Running those jobs on Batch before Dec 1 locks in the lower intro Batch rates ($1 / $5) while they last.
- Grounding: stay under the shared 5,000 free Search/Maps grounding requests/month when possible; overage is $14 / 1,000.
Unknowns (not published by Google on the pages we checked)
- Exact consumer Google AI Pro / Ultra monthly prices (not fetched for this article).
- Whether “Google AI Pro plan” on the API pricing page maps 1:1 to every Google One AI Pro SKU name in every country.
- Independent third-party benchmarks for Argon (none cited here).
- Priority / other inference tiers appear on the pricing page for some models; we limited cost math to Standard, Batch, and Flex as listed for
gemini-4-argon.
FAQ
Is Gemini 4 Argon free? Not on the Gemini API Free tier. Google’s pricing page states it is not available on the Free tier or on the Google AI Pro plan for API. Consumer app access for Ultra/Pro subscribers is described on the GA blog; that is not the same as free API usage.
How much does it cost after November 30? On the paid API tier, Standard becomes $4 per 1M input and $20 per 1M output (including thinking). Batch and Flex become $2 / $10. Context caching Standard becomes $0.20 (implicit caching only).
Can Google AI Pro subscribers use it? Per the October 13 blog, Google AI Ultra and Pro subscribers can use it in the Gemini app and AI Mode in Search. Per the API pricing page, it is not available on the Free tier or on the Google AI Pro plan for API. Attribute each claim to its source; we do not merge them into a single rule Google did not write.
Is it available in the API now? Yes, as of the October 13, 2026 GA post, for developers and enterprises via the API in AI Studio and the Gemini platform on the paid tier. Model ID: gemini-4-argon.
What is the context window? Google’s models page describes a unified 1M context window and long decodes up to 1M tokens; the GA blog describes a 1M unified input-output token limit via the Gemini API.
How we reviewed
We built this from Google’s official pages checked on October 16, 2026 (America/Sao_Paulo): the October 13 GA blog, the September 30 Fairwind announcement, the Gemini API pricing page, and the models list. Cost examples are our arithmetic on published Standard/Batch/Flex rates. We have not run Gemini 4 Argon ourselves.
Sources (official pages, checked on October 16, 2026 ~01:10–01:15 BRT)
- Google, Gemini 4 Argon generally available (Oct 13, 2026): blog.google/…/introducing-gemini-4-argon/ (HTTP 200)
- Google, original Fairwind announcement (Sep 30, 2026): blog.google/…/gemini-4-argon/ (HTTP 200)
- Gemini API pricing (
gemini-4-argon): ai.google.dev/gemini-api/docs/pricing (HTTP 200; page “Last updated 2026-10-15 UTC”) - Gemini API models: ai.google.dev/gemini-api/docs/models (HTTP 200; page “Last updated 2026-10-13 UTC”)
- Anthropic pricing (Opus 5.5 comparison row only): platform.claude.com/docs/en/about-claude/pricing
- OpenAI API pricing (GPT-6.1 Sol / GPT-6 Sol comparison rows): platform.openai.com/docs/pricing