Google unveils Gemini 4 Argon to reclaim AI frontier

Google Alphabet Inc. ↗ 1.62% on Wednesday announced Gemini 4 Argon, the first model in its next-generation Gemini 4 series, marking the company's most ambitious attempt to reclaim its position at the frontier of artificial intelligence after months of delays and internal setbacks.

The model, which Google is calling its "next era of frontier intelligence," is designed for deep reasoning across complex, long-running tasks in software engineering, cybersecurity, finance, and legal work. Access is initially limited to a small group of vetted cybersecurity partners through Google's Fairwind Program, with no public release date yet announced.axios+1

Benchmark Claims and Cybersecurity Focus

Google says Gemini 4 Argon outperforms rival frontier models on several key benchmarks. The company claims a score of 77.9% on DeepSWE v1.1, a test of real-world engineering tasks, compared to 74.2% for Anthropic's Claude Opus 5.5 and 74.1% for OpenAI's GPT-6 Astra. On CWE-bench v1, which measures the ability to find and patch security vulnerabilities, Argon ties with GPT-6 Astra at 68%.businessinsider+1

The model also boasts a 1-million-token output limit, up from 64,000, enabling far longer and more complex reasoning chains. "When the model has the headroom to think deeply and generate hundreds of thousands of tokens in a single trajectory, it adds a new level of depth in reasoning to solve tough problems in one go," Google said.9to5google

A Long Road to Launch

Gemini 4 Argon arrives nearly a year after Google's previous flagship generation, Gemini 3. The company had promised a new frontier model, Gemini 3.5 Pro, in June, but that release never materialized. Axios reported in July that poor morale among Google DeepMind employees contributed to the delay. Business Insider reported that the model was postponed multiple times internally due to poor performance.axios+1

"Argon is a well-rounded model that has frontier capabilities across several domains," Tulsee Doshi, head of Gemini products at Google DeepMind, told Axios.axios

Already at Work Inside Google

Internally, Google says thousands of employees are already using Argon for coding, research, and writing tasks. The company highlighted one case in which a team of Argon agents analyzed fleet-wide telemetry data to identify memory optimizations across Google's data centers, freeing up over 300 tebibytes of memory. Argon agents are also migrating C/C++ codebases to Rust at scale, including up to 800,000 lines for the Fuchsia OS Zircon kernel.9to5google

The broader rollout will begin with Google AI Ultra subscribers and paid API customers, though the company has not specified a timeline.9to5google