Gemini 4 Argon runs 33% cheaper than ChatGPT 6.1, cuts hallucinations to 15%

A Geeky Gadgets overview of Gemini 4 Argon highlights 3D modeling with X-ray and exploded views, app replication from screenshots and long-video comprehension. The model reportedly costs about 60% of rivals to run, posts a 15% hallucination rate on document analysis versus 51% for ChatGPT 6 Astra, and scores 91.7% on LFBench.

• Running cost is about 60% of ChatGPT 6.1 and Opus 5.5
• Document-analysis hallucination rate of 15% vs 51% for ChatGPT 6 Astra
• LFBench video score of 91.7%, turned a 3.5-hour lecture into a course
• Rebuilt the Toggle Track app from screenshots with near-identical features