As the first model in the Gemini 4 generation, Argon is designed for coding, business use and cyber defence. It will first be made available through the Fairwind programme before being rolled out to Ultra subscribers and via the API.
Google presented on September 30 Gemini 4 Argon, the first frontier model of his family Gemini 4. Designed for long and chained tasks — software development, business documentary work, IT defense — it is not yet open to the general public: Google gives it priority to trusted testers and selected cybersecurity teams via its program Fairwind, time to strengthen protection against malicious use and quick injection. Thousands of Google employees are already using it internally, including debugging and migrating code bases. Google AI Ultra subscribers and API paying customers must follow "nearly".
Top scores, an output that goes up to a million chips
Google claims 77.9% on DeepSWE v1.1, ahead of Claude Opus 5.5 (74.2%) and GPT-6 Astra (74.1%), first place on AutomationBench (51.3%), a lead on CWE-bench v1 (68%) and 91.7% on LVBench, a long video comprehension test. Another notable change: the exit limit climbs to one million tokens, compared to 64,000 before.
On the rate side, a launch period sets the price at 2 dollars per million tokens in entry and 10 dollars in exit, with 95% discount on cached entries; Then he'll move to $4 and $20.
Read the full article on the original source
Go to the source →