Running MiniMax H3 text-to-video locally on a 4GB RTX 3050 laptop
yushairiegalaxy96 · reddit · 2026-08-19
The author shares a local MiniMax H3 text-to-video setup on an RTX 3050 laptop (4GB VRAM, 16GB RAM): INT8/INT4 quantized pruned unet, Qwen3VL-32B NVFP4 clip, fp16 VAE, plus a LightX2V Turbo 8-step LoRA and the default T2V workflow. A 10-second 608x352 (0.2MP) 16:9 clip takes 687 seconds. They ask whether enabling Sage Attention, Comfy Kitchen, more steps, or a better unet/clip would help.
More from Infra
- Google's Willow chip cuts quantum errors 40x, raising Q-Day concerns for Bitcoin — jamestagg · 2026-10-16
- Vibe Coding a Data Pipeline Is Easy; Running It at Billions of Rows Is Not — Zachly · 2026-10-16
- Large data centers employ only ~50-150 permanent on-site staff, Microsoft says — brezina · 2026-10-16
- Prediction: Opus 5.5-level models on laptops by 2027, backed by Qwen 27B data — cephaloform · 2026-10-16
- Oracle down 12% in a week, nearly 50% off June high as OpenAI exposure spooks market — alvelda · 2026-10-16
- US data center moratoriums top 400 — an AI bull explains why they're actually good for AI — kevinsxu · 2026-10-16