Hands-on with Gemini 4 Argon: faster than Opus 5.5 but hallucinates more and costs more
prasenx · x · 2026-10-16
Developer prasenx spent a full day with Gemini 4 Argon and shared a candid take:
- A huge upgrade over past Gemini models and "scary fast" — notably quicker than Claude Opus 5.5
- But it hallucinates a lot, so he wouldn't trust it with real work
- Strong at 3D: with the right prompts and skills it excels at three.js and Blender (screenshots show iterative refinement of a 3D snow globe scene with precise texture/lighting edits)
- Expensive
- Verdict: for daily tasks and real coding workflows he still trusts Opus 5.5 and Grok more; a nice upgrade overall but not better than Opus 5.5
Related event: Hands-On: Gemini 4 Argon Blazing Fast but No Drop-in Replacement(2 posts)→
More from coding & agent
- vLLM Ships Day-0 Support for Meta's 816B Open-Weight Muse Spark Open — woosuk_k · 2026-10-16
- Same 3 Coding Packets, Two Agents: Claude Opus 5.5 Beats Gemini 3.8 Flash on Every One — umeshonai · 2026-10-16
- State machines are underrated and agents are great at writing them, says David Khourshid — TejasKumar_ · 2026-10-16
- Open-source Pi desktop client ships with session history, scheduled tasks, Codex-like UI — solyarisoftware · 2026-10-16
- ChatGPT brings computer use to Chrome: reads tabs and controls the browser — craigsdennis · 2026-10-16
- iMessage AI agent showdown: only Instinct fully completes a flower-buying task — msg · 2026-10-16