Google has launched Gemini 4 Argon, its first major model in over seven months, which has garnered attention for matching OpenAI’s GPT-6 Astra in independent testing. However, despite its competitive capabilities, Argon does not surpass Anthropic’s Claude Opus 5.5 in performance.
One of the notable challenges with Gemini 4 Argon is its token consumption; it utilizes more than twice as many tokens per task compared to Astra, which could impact its overall cost-effectiveness. While the per-token price is relatively low, the increased token usage may limit its appeal for certain applications.
Currently, select testers are granted access to Argon, with broader availability through API and paid tiers expected to follow. This release highlights Google’s ongoing efforts to remain competitive in the rapidly evolving AI landscape.
Compiled automatically by the Tech AI Newsdesk from public AI-news sources and summarised in our own words.
