Google Unveils Gemini 4 Argon with 1 Million Token Limit

Google introduced the new Gemini 4 Argon language model, marking its first major release since February of last year. This development replaces Gemini 3.1 Pro and is positioned as a breakthrough in long-form text generation.

The model outperforms competitors Claude Opus 5.5 and GPT-6 Astra in benchmark tests. Its key feature is a limit of 1 million output tokens, enabling the creation of large-scale projects from a single prompt.

Key features of Google's new model:

• High performance test scores
• Output limit up to 1 million tokens
• Code generation capabilities for large projects

Access to the tools is now open to developers. The model is integrated into the AI Studio platform and Google's AI agent Antigravity.