Google has announced its new frontier model, Gemini 4 Argon, which is currently rolling out to trusted cyber defenders through the Fairwind Program. Built to sustain deep reasoning across complex, long-horizon workflows, Argon delivers frontier performance in real-world software engineering, enterprise knowledge work, and cybersecurity defense.
The model will launch at an introductory price of $2 per million input tokens and $10 per million output tokens, with cached input tokens discounted by 95%. To support long-horizon use cases, Google has significantly expanded the output token limit to an industry-leading 1 million tokens, up from the previous 64K.
Internal Applications and Workflow Enhancements
Gemini 4 Argon is already powering internal workflows at Google, accelerating breakthroughs across specialized tasks:
- Quantum Algorithmic Optimization: Argon helped quantum computing researchers beat a published baseline by 40% in minutes by optimizing spacetime resources.
- Memory Efficiency: Argon agents analyzed fleet-wide profiling telemetry to autonomously apply memory optimizations across Google’s data centers, freeing up over 300 TiB of memory with projected total savings up to 1 PiB.
- Codebase Migrations: Agents are actively scaling C/C++ to Rust migrations from tens of thousands of lines up to over 800K lines for the Fuchsia Zircon kernel. For open-source video decoder libgav1, Argon replaced 32K lines of SIMD code with safe, compiler-vectorized Rust, running 2.7x faster than the initial Rust port.
Leading in Defensive Cybersecurity
Trained to autonomously find, validate, and patch critical software vulnerabilities, Gemini 4 Argon is being released to trusted defenders and internal teams without cyber guardrails to leverage its full capabilities. Partners like Wiz are already utilizing Argon through the Scan for Good initiative, where it successfully uncovered a critical vulnerability in global hospital software that previous frontier models missed. On CWE-bench v1, Argon ties for first place with a top score of 68%.
Robust Frontier Safeguards
Before broad commercial availability to developers, enterprises, and consumers, Google is enforcing a phased rollout and strengthening safeguards across four core areas:
- Misuse Prevention: Designed to refuse harmful requests while permitting dual-use scientific research in line with the Frontier Safety Framework, backed by internal activation monitoring.
- Prompt Injection Resilience: Utilizing automated red teaming and adversarial training, Argon leads in robustness on Gray Swan’s Indirect Prompt Injection benchmark.
- Misalignment Monitoring: Deploying mitigations that monitor chain-of-thought and actions to halt execution when necessary without compromising transparency.
- System Hardening: Isolating and sealing sandboxed environments prior to high-risk evaluations and training runs.
