Google launches Gemini 4 Argon globally with support for up to 1 million tokens

Announcement and availability

Google CEO Sundar Pichai announced on Tuesday, October 13, that the Gemini 4 Argon model is now available worldwide. The release is designed for developers, businesses, and subscribers of the Pro and Ultra plans, offering varying levels of access depending on the product.

Users of the Gemini app and Google Search on Pro and Ultra plans can already access the new model. Google Antigravity offers this feature exclusively to Ultra subscribers. Developers and businesses can access it through the paid programming interface and the Gemini platform.

Gemini 4 Argon is here. It’s available globally today for developers and enterprises in the paid tier of the Gemini API and the Gemini platform, for Ultra subscribers in Google Antigravity, and Ultra and Pro subscribers in the Gemini app and Search. Hope you all enjoy it! https://t.co/2jyAgqoylU

— Sundar Pichai (@sundarpichai) October 13, 2026

Extended generation and token limit

Argon supports extended generation of up to 1 million tokens, using a shared window of 1,048,576 tokens for input, reasoning, and output. The default API limit for reasoning and output is 65,536 tokens; to generate longer responses, extended generation configuration, known as long decode, is required.

Google highlighted use cases such as financial research, drafting legal documents, and modifying large software projects. Internal teams at the company already use Argon for programming, research, and writing tasks. During code migration, the company conducts automated tests, audits, and human reviews before implementation in production.

Performance in comparative tests

In the updated Vals Index as of October 12, Argon leads with 68.90%, excelling in the Finance Agent v2 evaluation. However, other systems outperform Argon in specific tasks, such as code migration and legal research. In DeepSWE v1.1, Argon achieved 77.9% compared to 74.1% for GPT-6 Astra, while in FrontierSWE v2, it scored 55%, below the competitor’s 65.5%.

Pricing and access plans

The introductory API pricing is $2 per million input tokens and $10 per million output tokens, including reasoning tokens. This offer is valid until November 30, 2026. From December 1, prices will increase to $4 and $20, respectively. Billing is based on the volume of tokens processed and is not tied to the monthly fees of the Pro or Ultra plans.

Frequently Asked Questions

• Who can use Gemini 4 Argon? Pro and Ultra subscribers in the Gemini app and Google Search; Ultra subscribers in Antigravity; and developers and businesses using the paid API tier and the Gemini platform.
• Is the release global? Yes.
• What does the 1 million token limit mean? Extended generation can reach this limit, but input, reasoning, and output share a window of 1,048,576 tokens. The default API limit is 65,536 tokens for reasoning and output; exceeding this requires specific configuration.
• Does Argon lead all tests? No. While it leads in the consulted Vals Index, other models outperform it in specific components. In some evaluations, Argon trails behind competitors.