Voxel diorama comparison: User reports similar completion times but higher V4.1 token use
A user said V4-Flash-Vision-Exp and V4.1-Intermediate both produced videos alongside their voxel dioramas, but later reported a video-capture failure in another V4.1 run.
TLDR
A user comparing V4-Flash-Vision-Exp and V4.1-Intermediate on a voxel diorama task reported similar completion times. They said V4.1 used 25 million input tokens and 215,000 output tokens, versus 9.2 million input and 119,000 output tokens for V4-Flash-Vision-Exp. Both runs also produced videos, according to the user. In a later reply, they reported that another V4.1 run had botched video capture.
Combined views
10.4K
2 Sources, first seen 23d ago