Gemini 4 Argon: our next era of frontier intelligence

Google has announced Gemini 4 Argon, its new frontier AI model, initially rolling out to trusted cyber defenders through the Fairwind Program. The model delivers frontier performance across real-world software engineering, enterprise knowledge work in legal and finance, and cybersecurity defense. Internally, Argon has powered significant advancements including quantum algorithmic optimization, fleet-wide memory efficiency improvements freeing over 300 TiB, and large-scale C/C++ to Rust codebase migrations. The model features an industry-leading 1M output token limit (up from 64K) and achieves state-of-the-art results on benchmarks like DeepSWE v1.1 (77.9%), Vals Index, AutomationBench (51.3%), and CWE-bench v1 (68%). For cybersecurity, Argon is released without guardrails to trusted defenders and has already uncovered critical vulnerabilities in healthcare software through Wiz's Scan for Good initiative. Before broad availability, Google is strengthening safeguards across four areas: defending against misuse (including CBRN attacks), prompt injection resilience, misalignment monitoring, and hardened sandboxed environments. Argon will launch at $2 per million input tokens and $10 per million output tokens, with rollout beginning to paid API customers and Google AI Ultra subscribers.

(Source：Gemini)