Google on Tuesday made Gemini 4 Argon, its new flagship model, available to developers on the paid tier of the Gemini API through Google AI Studio, as well as in other developer and consumer tools. The rollout introduces a unified 1 million token input and output budget and introductory pricing that undercuts several rival flagship models.
Google has released Gemini 4 Argon, its newest flagship model, to developers on the paid tier of the Gemini API through Google AI Studio, along with several of the company's other developer and consumer tools.
The company first announced Argon on September 30, when external access was limited to selected cyber defenders in its Fairwind Program and other trusted testers, with Google employees also using it internally. In a post on X, Google's Logan Kilpatrick said the model had been 'battle tested by over 100,000 people at Google over the last month.'
Developers can call the model as gemini-4-argon through Google's Interactions API, with support in the google-genai SDKs. Android developers can also build it into apps through Firebase AI Logic.
Pricing
During the introductory period, which runs through November 30, developers pay $2 per million input tokens and $10 per million output tokens, the same price as OpenAI's GPT-6.1 Sol. After that date, the model will cost $4 and $20 per million tokens, matching Anthropic's Claude Opus 5.5. Cached input tokens are 95 percent off during the introductory period.
At the introductory rate, Argon also costs a fifth of what Anthropic charges for Claude Fable 5.1 and OpenAI charges for GPT-6 Astra, which both list at $10 and $50 per million tokens.
Token limits
Google now describes Argon's limit as a unified 1 million token budget covering both input and output. That differs from the original September 30 announcement, which described a 1 million token output cap, up from 64,000 in earlier Gemini models.
Because the budget is shared, a long prompt can reduce the room a model has to respond. A request loading 800,000 tokens of code, for example, leaves 200,000 tokens for the response. By comparison, Claude Opus 5.5 combines a standard 1 million token context window with a separate output cap of 128,000 tokens.
Google says the extra room will let Argon keep working on long, multistep coding, science and agent jobs, and that the model ranks first on the Vals Index.
Analysis
Why This Matters
- Google's flagship model is now available to paying developers at prices that undercut several competing top tier models.
- The unified 1 million token input and output budget changes how long context workloads are priced and constrained, and differs from rivals that keep output caps separate.
- The launch intensifies price competition among Google, OpenAI and Anthropic in the high end of the AI model market.
Background
Gemini 4 Argon is Google's newest flagship model in the Gemini line, first announced on September 30. Its initial access was narrow: selected cyber defenders in the Fairwind Program, other trusted testers and Google employees. Tuesday's rollout opens the model to developers through the paid tier of the Gemini API, Google AI Studio and other tools.
Argon now enters a competitive field in which OpenAI's GPT-6 Astra and Anthropic's Claude Fable 5.1 both list at $10 and $50 per million tokens. Google's introductory pricing is set well below that, before settling at the same rate as Claude Opus 5.5 after November 30.
Key Perspectives
Google: Argon is a flagship model that has been battle tested by more than 100,000 employees. Google says the unified million token budget supports long, multistep coding, science and agent workloads, and it points to a first place ranking on the Vals Index.
Developers: Introductory pricing is attractive, particularly compared with $10 and $50 per million token models. However, the shared budget means long inputs cut into output capacity, so teams need to plan prompts with that tradeoff in mind.
Competitors: OpenAI and Anthropic face price pressure. Google matches OpenAI's GPT-6.1 Sol on price during the promo and matches Anthropic's Claude Opus 5.5 after it, while undercutting GPT-6 Astra and Claude Fable 5.1 in the meantime.
Critics/Skeptics: The unified token budget is a tradeoff rather than a pure upgrade. An 800,000 token input leaves only 200,000 tokens for output. Vendor benchmark claims are also generally treated with caution until independently verified.
What to Watch
- How Argon performs in practice on long coding, science and agent tasks, where the shared input and output budget is most consequential.
- Whether introductory pricing is extended beyond November 30, or whether rivals respond with price cuts of their own.
- Further consumer and developer rollouts, including the Gemini app tiers and Antigravity subscribers mentioned in Google's rollout post.