Google claims Argon delivers superior reasoning while curbing the repetitive hedging common in previous iterations. Internal testing already supports this, with the company utilizing the model to patch security vulnerabilities and optimize the computing infrastructure behind YouTube comments and Wing’s drone physics simulators. By positioning the model at $2 per million input tokens, Google is aggressively undercutting the pricing structures of GPT-6 Astra and Claude Fable 5.1.
Google Challenges AI Rivals With Gemini 4 Argon Launch
Priced at a fraction of its competitors, Google’s new Gemini 4 Argon model arrives to confront the dominance of OpenAI and Anthropic. The release signals a pivot for the search giant, moving from months of relative silence to a direct assault on the current benchmarks in coding, finance, and cybersecurity.
Access is currently restricted to Pro and Ultra subscribers via the Gemini app, Search, and the Antigravity development platform. While an introductory pricing period ends in November, the model is expected to remain cost-efficient compared to industry peers. This release follows a controlled testing phase with cybersecurity experts, marking Google’s first significant move in the sector since the late 2025 launch of Gemini 3. Coming on the heels of Meta’s Muse Wonder Full announcement, Argon reinforces the volatility of the current market and challenges the assumption that the frontier AI race is limited to two major players.
Comments (0)
No comments yet. Be the first!