Google announced Gemini 4 Argon, a frontier model released first to trusted cyber defenders

On September 30, 2026, Google announced Gemini 4 Argon, a frontier model for long-horizon reasoning that began rolling out to trusted cyber defenders through the Fairwind Program, with a 1M-token output limit.

Archive entry: this covers an event from September 30, 2026, and we published it on November 8, 2026. The details come from sources dated at the time and may have changed since.

On September 30, 2026, Google announced Gemini 4 Argon, a frontier model built for long-horizon reasoning, in Google's announcement of Gemini 4 Argon on The Keyword. It began rolling out to trusted cyber defenders through the Fairwind Program, and Google said it was taking part in the US government's voluntary process for pre-release model access.

What Google reported

The output token limit rose to 1M tokens, up from 64K.
Google said Argon set the state of the art on DeepSWE v1.1 with 77.9%. It also led the Vals Index, ranked #1 on Zapier's AutomationBench with 51.3% and scored 91.7% on LVBench, and it tied for first on CWE-bench v1 with 68%.
Argon agents freed over 300 TiB of memory across Google data centers, with estimated total savings of 500 TiB to 1 PiB, and were migrating C/C++ codebases to Rust, including more than 800K lines for the Fuchsia Zircon kernel.
For libgav1, Argon replaced 32K lines of SIMD code in a Rust port, producing a memory-safe decoder that ran 2.7x faster than that port. In one quantum-computing example it beat the published baseline by 40%.

What it meant for people who build with AI

Argon was the first Gemini 4 model. Google released it in stages, starting with security defenders, and its 1M-token output limit left room for very long agent runs.

Access and pricing

Trusted defenders and internal Google teams got Argon without cyber guardrails, and the wider release was to start with paid API customers and Google AI Ultra subscribers. The introductory price was $2 per 1M input tokens and $10 per 1M output tokens, with cached input at 95% off. After the introductory period, the price was to be $4 per 1M input tokens and $20 per 1M output tokens.

Sources

Gemini 4 Argon: our next era of frontier intelligence (Google (The Keyword), September 30, 2026, official)

Written with AI assistance, strictly from the sources listed above. Before it went live, automated checks verified its dates, sources and every language edition.