Gemini 4 Argon in Cursor rivals Opus 5.5 but feels slower, developer says
A developer testing Gemini 4 Argon in Cursor says it is hard to distinguish from their daily Opus 5.5 on scoped tasks, though it feels slower, has a questionable hallucination rate, and is not in the Antigravity Pro tier.
first thoughts on Gemini 4 Argon via Cursor its definitely a very capable model and I'm having a hard time differentiating it from my daily driver Opus 5.5 on tasks I've mostly scoped out already it feels a bit slower I've been using it effectively to gather and synthesize context across a lot of different threads I really appreciate its writing style, feels very distinct and not Claude or GPT coded I may use it in planning sessions as another review agent for model diversity. Theo has a post showing it still has an absurdly high hallucination rate, which conflicts with its AA benchmark. Makes me question it's actual trustanility it's a pity it's not in the Antigravity Pro tier otherwise I'd probably use it a decent amount in my personal life. for now, it