GPT-6.2 Sol voxel benchmark: mostly a regression vs 6.1, but much faster and cheaper
Angaisb_ · x · 2026-11-18
A side-by-side voxel generation comparison of GPT-6.2 Sol vs GPT-6.1 Sol across 32³, 64³ and 128³ resolutions and 15 prompts (penguin, scooter, carousel, tree frog, diving helmet, Mars rover, etc.) finds most outputs got worse in 6.2, with a few improvements — a regression in voxel quality. The upside: 6.2 is significantly faster, with most samples completing in 30–130s at $0.03–$0.10 per run.
More from Models
- Karpathy: Google's undistilled Gemini Carbon thinks differently, and it's the only full-stack AI company — beffjezos · 2026-11-18
- SpaceX CFO hails Grok Bot as 'probably the best agentic AI solution out there' — XFreeze · 2026-11-18
- Sam Altman admits new OpenAI models are smart but hard to work with, vows to fix UX — pvncher · 2026-11-18
- Meta FAIR & NYU controlled study: visual prior unnecessary for multimodal models, dedicated capacity is key — NielsRogge · 2026-11-18
- GPT-6.2 Astra scores 20 points above prior best on DeepsecBench security tasks — cramforce · 2026-11-18
- GPT-6.2 Astra tops AA-3D, winning 139 of 152 human preference votes — ArtificialAnlys · 2026-11-18