Google unveils Gemini 4 Argon for long-horizon work, starting with trusted cyber defenders

Google announced Gemini 4 Argon on September 30, 2026, positioning it for complex, long-horizon workflows spanning software engineering, enterprise knowledge work such as legal and finance, and cybersecurity defense. The company says the model is rolling out to a group of trusted cyber defenders through its Fairwind Program. Before widening access, Google says it will continue strengthening safeguards and collecting feedback from early testers.

The announced introductory API rates are $2 per million input tokens and $10 per million output tokens, with cached input tokens priced 95% below the input-token rate. Google says the rates will rise after the introductory period to $4 per million input tokens and $20 per million output tokens. Argon’s output limit is 1 million tokens, up from 64,000, which Google presents as support for longer reasoning and multi-step tasks.

Cybersecurity is central both to the model’s stated use cases and its initial rollout. Google says Argon can autonomously find, validate and patch software vulnerabilities, and that trusted defenders and Google’s internal teams will receive it without cyber guardrails. The company reports a tied first-place score of 68% on CWE-bench v1, saying this builds on Gemini 3.8 Flash Cyber’s performance on the earlier CWE-bench v0.

Broader availability is planned in stages. Google says it will begin expanding access to paid API customers and Google AI Ultra subscribers after further safeguard work and early testing. The announcement therefore sets out not only the model’s capabilities, but also an initial deployment path that puts trusted cyber defenders ahead of broader customer access.