GLM 5.3 Flash vs DeepSeek V4.1: voxel scene test ends in a draw despite 1.5x more steps
teortaxesTex · x · 2026-09-08
A user tested GLM 5.3 Flash against DeepSeek V4.1 on identical voxel diorama tasks with similar completion times. DeepSeek had fog and distance issues when recording video, but the author judges its underlying model better overall — calling them roughly equal.
Details: V4.1 used 1.5x more steps (150), with 25M tokens in and 215K out; V4-Flash-Vision-Exp used 9.2M in and 119K out. Both models produced their own videos, and another V4.1 run (12M in, 168K out) was still processing.
More from Models
- Claude subreddit floods with complaints about new Opus; users downgrade to Opus 4.6 — kimmonismus · 2026-09-08
- LinkedIn Is LLMs' #2 Cited Site — Your Posts Are Becoming AI Knowledge — jaindl · 2026-09-08
- Benchmaxxing exposed: fresh benchmark drops model scores from 89% to 19% — Able-Line2683 · 2026-09-08
- Dev prefers Fable 5.1/5.6 Sol for planning and coding, eyes Grok 4.7 next — iannuttall · 2026-09-08
- Minimax H3 speedup modes strip realism, full steps needed for production — Choiced_Gamer · 2026-09-08
- User burns 40% of weekly GPT 6 Astra limit in 11 hours, complaining quotas drain fast — iScienceLuvr · 2026-09-08