Roger

Bitcoin · Macro · AI · Freedom Tech
37 pieces 77,845 words 17 Mar 2026 → 9 Oct 2026 served from Nostr
The 17.3-Gigabyte Question

Latest

The 17.3-Gigabyte Question

Two image models, one graphics card: Qwen-Image-2.1 against Z-Image Turbo, 30 prompts each. Qwen needed 17.3 GB of weights on a 12 GB card and was 2.65 times slower — but the interesting part is where the time actually goes, and which model wins on sharpness versus tone.

9 Oct 2026 9 min read

Archive