Google Gemini 4 Argon Targets Frontier AI Leadership With Coding and Cybersecurity Gains

Google promised to introduce Gemini 3.5 Pro in June, but spent the summer unveiling a series of smaller Flash models. Now, Google is ready to take on the frontier AI market again with Gemini 4 Argon.

Google claims the new AI model delivers industry-leading performance in coding, knowledge work, and cybersecurity. However, Gemini 4 Argon has not yet been cleared for general use.

Google Is Already Using Gemini 4 Argon Internally

While most users will have to wait to test Gemini 4 Argon, Google says its engineers are already using the model extensively. According to Google, Argon analyzed “fleet-wide telemetry data” that helped the company save 300 TiB of memory across its data centers.

The Gemini 4 Argon agent is also helping migrate Google’s C/C++ codebase to Rust. The work includes thousands of lines from the core re2 and libgav1 libraries, as well as more than 800,000 lines of the Fuchsia OS Zircon kernel.

Gemini 4 Argon Benchmark Results

Google has released several benchmarks to support its performance claims. In the Software Engineering DeepSWE v1.1 benchmark, Gemini 4 Argon scores 77.9 percent—higher than GPT-6 Astra, Fable 5.1, and Opus 5.5.

Google also points to Argon’s industry-leading score in the Vals Index test of economic analysis and promises similar capabilities across a range of long-term tasks.

Gemini 4 Argon API Pricing and Token Limits

Gemini 4 Argon remains in limited testing, but Google has announced API pricing. For a limited time, Argon will cost $2 per million input tokens and $10 per million output tokens, with a 95 percent discount on cached input tokens.

Google also confirmed that Gemini 4 Argon will support an output limit of up to 1 million tokens. That is a major increase from the previous Gemini model’s 64,000-token limit. Google says the larger context will allow users to complete more difficult tasks in a single step.