Google has introduced Gemini 4 Argon, a new frontier AI model that, according to The Decoder, matches the performance of OpenAI's GPT-6 Astra in independent testing. However, it still trails behind Anthropic’s Claude Opus 5.5, which currently leads in this category.

The Decoder also notes that while Gemini 4 Argon offers a low per-token price, it requires more than twice as many tokens per task compared to GPT-6 Astra. Select testers have been given early access to the model, with API availability and paid tiers expected to follow over the coming months.

This development comes as Japanese markets closely watch AI advancements, given their growing impact on technology stocks and automated trading strategies. The competition among leading AI models continues to influence investor sentiment in the region’s tech sector.