Google introduces Gemini 4 Argon, with access starting among cyber defenders
Google’s newest frontier model has a one-million-token limit, but its first rollout is restricted while the company tests safeguards.
Google announced Gemini 4 Argon on 30 September 2026. The model is built for long-running software engineering, enterprise knowledge work and defensive cybersecurity, but it is not a general public release yet: initial access is going to selected cyber defenders through Google DeepMind’s Fairwind Program.
Argon has a one-million-token limit and introductory API pricing of $2 per million input tokens and $10 per million output tokens, with cached input priced at 5% of the normal input rate. Google says it plans to expand access to developers, enterprises and consumers after more testing, without giving a release date.
Google reports 77.9% on DeepSWE v1.1 and 51.3% on AutomationBench. Those are vendor-reported launch figures. Artificial Analysis separately scores Gemini 4 Argon at 53 on Intelligence Index v4.3.2, level with GPT-6 Astra and Claude Fable 5.1 in that snapshot.
Gemini 4 Argon is a hosted proprietary model. Google has not published downloadable weights, so it does not belong in this site’s local-model compatibility catalogue.
Sources
This is a summary of reporting published elsewhere. Where a claim matters, follow the source — it is the record, this is not.
Which models can your PC run?
Scan your hardware and see every model that fits, with the memory each one needs.