Google has announced Gemini 4 Argon, its latest frontier AI model designed for high-performance coding, knowledge work, and cybersecurity applications. Internal Google engineers are currently utilizing the model, which reportedly helped save 300 TiB of data center memory and migrated over 800,000 lines of code in the Fuchsia OS Zircon kernel to Rust, according to the company.
External availability is initially limited to select partners in Google’s Fairwind Program to focus on cyberdefense testing. Security firm Wiz has already used Argon to identify a critical hospital system vulnerability. Google stated that Argon incorporates reasoning transparency mechanisms that monitor chain-of-thought processes to halt potential out-of-bounds behavior during execution.
Google announced introductory API pricing of $2 per million input tokens and $10 per million output tokens, along with a 95 percent discount for cached inputs. The model expands the output limit to 1 million tokens, up from 64,000 in prior models. General availability will eventually roll out to paid API users and Google AI Ultra subscribers.
Why it matters
A 1-million-token output limit enables models to execute longer reasoning chains and complex tasks in single steps.
Phased rollouts focused on cyberdefense reflect heightened industry caution around frontier model misuse and safety risks.
Introductory API pricing signals aggressive competition on developer costs among top-tier foundation model providers.
Source: arstechnica.com


