How much will Gemini 4 Argon cost?

Google has published introductory API pricing: $2.00 per 1M input tokens and $10.00 per 1M output tokens, rising to standard rates of $4.00 and $20.00 per 1M tokens. Prompt caching offers a 95% discount on cached inputs.

Who can use Gemini 4 Argon right now?

Access is currently gated to select cyber defenders under Google's Fairwind Program and internal DeepMind teams. References and backend integrations have surfaced in Google Cloud Vertex AI, suggesting broader access is coming.

What can the 1 million token output window do?

It allows the model to write entire multi-file software projects, refactor complete enterprise codebases, and draft exhaustive technical documentation in one unbroken completion pass.

Large Language ModelsTop Companies' AI MovesAI Business & IndustryTop Companies AI5h agoNov 2, 2026, 06:30 JST

Gemini 4 Argon hits Vertex AI staging, launch near

Send on LINE

3 Key Points

What happened
Gemini 4 Argon, capable of 1 million tokens of continuous output, has appeared in Google Cloud Vertex AI staging after being limited to Fairwind Program cyber defenders and internal DeepMind teams.
Why it matters
The staging appearance marks the second of Google's three-stage deployment steps, and history suggests broad availability follows quickly, which would give enterprise developers access to a model built for long multi-step tasks.

WHO IT HITSEnterprise developers and engineering teams building automated coding pipelines on Google Cloud stand to gain access to a much larger output window, while teams still using older Gemini endpoints may need to plan for deprecations.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, every morning.

Get these free every morning

Context & Analysis

Gemini 4 Argon was announced on September 30 as a leap in long-horizon reasoning and complex software engineering, but access was initially locked to select cyber defenders under the Fairwind Program and internal DeepMind teams. The model's appearance in Google Cloud Vertex AI staging indicates the second stage of Google's three-stage deployment sequence is now active.

Google's pattern for frontier models runs from gated specialist testing to cloud staging to general cloud and developer access. With Argon now showing up in Google Cloud's enterprise environment, the company has already published introductory API pricing, and older model iterations are being phased out to steer workload demand toward specialized endpoints.

Early real-world testing saw cybersecurity firm Wiz use Argon to map attack surfaces and discover a previously unknown critical vulnerability in global healthcare software, and Google DeepMind benchmarks show it outperforming competitor frontier models on multi-file engineering and complex reasoning tasks. Argon is expected to become the new default foundation for complex agentic tasks on Vertex AI.

FAQ

How much will Gemini 4 Argon cost?: Google has published introductory API pricing: $2.00 per 1M input tokens and $10.00 per 1M output tokens, rising to standard rates of $4.00 and $20.00 per 1M tokens. Prompt caching offers a 95% discount on cached inputs.
Who can use Gemini 4 Argon right now?: Access is currently gated to select cyber defenders under Google's Fairwind Program and internal DeepMind teams. References and backend integrations have surfaced in Google Cloud Vertex AI, suggesting broader access is coming.
What can the 1 million token output window do?: It allows the model to write entire multi-file software projects, refactor complete enterprise codebases, and draft exhaustive technical documentation in one unbroken completion pass.

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Get it free

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.