Google’s Gemini 4 Argon Goes Global at $2 per Million Input Tokens, but the Intro Price Doubles After November 30

Google made Gemini 4 Argon generally available worldwide on October 13, 2026, two weeks after it first went to a small group of cyber defenders. Paid Gemini API users can call it today at an introductory $2 per million input tokens and $10 per million output tokens. That price doubles to $4 and $20 on December 1.

Free Gemini users do not get it. Neither does the free tier of the Gemini API.

What Google launched today

Google CEO Sundar Pichai announced the release on X at 9:34 PM IST. The details sit in a post on The Keyword by Koray Kavukcuoglu, SVP at Google DeepMind and Google’s Chief AI Architect.

“Gemini 4 Argon is now globally available today for developers and enterprises via the API in AI Studio and Gemini platform, for Google AI Ultra subscribers in Google Antigravity, and Google AI Ultra and Pro subscribers in the Gemini app and AI Mode in Search,” Kavukcuoglu wrote.

Google first announced Argon on September 30, but only for partners in its Fairwind cyber defense program. Today is the first public release. Google calls it the first member of the Gemini 4 family and says more models will follow.

Who gets Gemini 4 Argon, product by product

Access depends on what you pay for. Here is the split, based on Google’s launch post, its developer post and its Gemini app post.

• Developers: the paid tier of the Gemini API, in AI Studio and Android Studio. The API model ID is gemini-4-argon.
• Google AI Ultra subscribers: Google Antigravity, AI Studio, the Gemini app and AI Mode in Search.
• Google AI Pro subscribers: the Gemini app and AI Mode in Search only.
• Enterprises: Gemini for business, the API and Antigravity.
• Free users: nothing. Since October 9, free Gemini users only get Flash-Lite, and Argon does not change that.

Google is also offering 10,000 scientists free access to Antigravity’s highest usage tier for six months, subject to selection.

Pricing: cheap now, double later

The Gemini API pricing page lists Argon at $2.00 per million input tokens and $10.00 per million output tokens through November 30, 2026. Output includes thinking tokens. From December 1 it becomes $4.00 and $20.00.

Cached input costs $0.10 per million now and $0.20 later. Only implicit caching works; explicit caching is not supported. Batch and Flex modes cost half the standard rate, and Priority costs $3.60 input and $18.00 output during the intro period.

Unlike Gemini 3.1 Pro, Argon has no separate higher price for long prompts. The rate is the same across the full window.

How Argon compares on price

All rival prices come from each company’s own pricing pages, in USD per million tokens at standard rates.

Model	Input	Output	Long prompt surcharge
Gemini 4 Argon (to Nov 30)	$2.00	$10.00	None listed
Gemini 4 Argon (from Dec 1)	$4.00	$20.00	None listed
OpenAI GPT-6 Sol	$2.00	$10.00	$4 / $15 above 272K input
Claude Opus 5.5	$4.00	$20.00	None listed
Meta Muse Wonder	$4.00	$20.00	See Meta’s terms

The pattern is obvious. Argon’s intro price matches GPT-6 Sol exactly, and its permanent price matches Claude Opus 5.5 and Muse Wonder exactly. Google is buying seven weeks of switching time.

Specs: a 1M token budget shared by input and output

According to the Google DeepMind model card, Argon takes text, images, audio and video, and outputs text. Its limit is “a combined input-output token context window of up to 1M.”

That detail matters. The 1M tokens are one shared budget, so a 900K token prompt leaves about 100K for the answer. Google calls it an “industry-leading 1M unified input-output token limit” and points developers to a new continuation protocol for long jobs.

The model card describes Argon as a new sparse mixture-of-experts model, trained from scratch on TPUs, not a fine-tune of an older Gemini. Its knowledge cutoff is March 2026.

The benchmarks, as Google reports them

Every number below is vendor reported. Google ran the tests and picked the comparisons. Argon also was not on Scale AI’s Humanity’s Sixth Sense leaderboard when we covered it, so independent results are still thin.

Google says Argon ranks first on the Vals Index at 68.9%, ahead of Claude Opus 5.5 at 67.0% and GPT-6 Astra at 63.1%. On Harvey’s Legal Agent Benchmark it scores 19.6%, against 5.4% for GPT-6 Astra and 3.8% for Opus 5.5.

Long context is its strongest area. On GraphWalks between 256K and 1M tokens, Google reports 84.2% for Argon versus 71.8% for GPT-6 Astra.

Google’s own table also shows where Argon loses. On FrontierSWE v2 it scores 55.0% against 65.5% for GPT-6 Astra. On Terminal-bench 4.0 it scores 57.4% against 66.4% for Claude Opus 5.5. “Frontier performance” in coding is not a clean sweep.

Safety: Google cannot rule out two critical thresholds

The model card is unusually frank. Under its Frontier Safety Framework, Google says it “cannot rule out” that Argon has reached the critical capability level for biological risk and for cybersecurity.

In both cases Google says it built a safety case and deployed safeguards that it assesses bring the risk below those levels. It also says it cannot rule out the “Stealth and Situational Awareness” threshold after Argon passed 14 of 15 challenges.

The launch post claims external bio red teamers unlocked less than 1% of tested sub tasks, and that Argon cut prompt injection attack success to 0.7% on the Gray Swan benchmark. The full Safety and Security Report was listed as “coming soon” when we checked.

Confirmed vs not confirmed

Claim	Status
Global launch on Oct 13, 2026	Confirmed by Google
API model ID gemini-4-argon	Confirmed on the pricing page
$2 / $10 until Nov 30, then $4 / $20	Confirmed
1M combined input and output tokens	Confirmed in the model card
Free Gemini app or free API tier access	Not offered
Separate max output cap below 1M	Not published
Independent benchmark results	Not yet available
Full Safety and Security Report	Announced, not yet seen
India specific pricing or limits	Not announced

What it means for Indian developers

Google says the rollout is global, and it lists no country exclusions, so Indian developers on a paid Gemini API account should see Argon now. Billing is in USD at the same rates.

The intro window is the practical point. If you run long document or codebase jobs, test them before November 30, while output costs $10 per million instead of $20. Then decide whether the results justify the full price.

Students and hobbyists relying on the free API tier or free Gemini app are locked out. A Google AI Pro plan gets you Argon in the app, but not in Antigravity.

Differences from Pichai’s post

Pichai’s post is accurate but compressed. “Search” means AI Mode in Search, not regular results. Pro subscribers get the app and AI Mode, but Antigravity is Ultra only. And the post does not mention that the $2 and $10 pricing is temporary.

FAQ

Is Gemini 4 Argon free?

No. It needs a paid Gemini API account, Google AI Pro or Ultra, or a Gemini for business plan. The free API tier lists it as not available.

How much does Gemini 4 Argon cost?

$2 per million input tokens and $10 per million output tokens until November 30, 2026. From December 1 it costs $4 and $20.

What is the Gemini 4 Argon context window?

Up to 1M tokens combined across input and output, according to Google DeepMind’s model card.

Is Gemini 4 Argon available in India?

Google describes the launch as global and lists no exclusions. Indian users need a paid plan or paid API access.

Are Google’s benchmark claims independently verified?

No. All benchmark figures in this article come from Google, and Google’s own table shows Argon trailing rivals on some coding tests.