Running MiniMax H3 locally: 2:19 video takes 5 hours on an RTX 3060
irmemon225 · reddit · 2026-08-21
A Reddit user tested MiniMax H3 R2V hybrid model for local video generation on an RTX 3060 12GB with 16GB RAM: each 10-second 0.8MP segment takes 20 minutes, and the full 2:19 video (14 prompts stitched together) took about 5 hours total.
Setup: int8 quantized hybrid model (minimaxh3hybridfl2varef2vab30-49-int8) plus Kijai's Turbo LoRA (lightx2v turbo 4step), default workflow with Sage Attention on, ersde sampler with beta, 6 steps; character images made with Anima.
More from Infra
- Agentic CPU Guidebook: Why Agent Workloads Point to a $221B Server CPU Market by 2030 — BenBajarin · 2026-10-16
- Haight Street Compute Market: a one-day trading floor for GPU deals in SF — ctjlewis · 2026-10-16
- Temporal Serverless Workers on Cloud Run: queue-driven autoscaling, scale to zero — rseroter · 2026-10-16
- LMA brings day-0 model support to TPU for the first time, spanning NVIDIA, AMD, Intel via SGLang — ying11231 · 2026-10-16
- Semtech's analyst-day slide cleanly explains the four signal-integrity impairments — BenBajarin · 2026-10-16
- Fortytwo Engine boosts Qwen3.8-27B accuracy 7% and cuts tokens 25% via Swarm Inference — rohanpaul_ai · 2026-10-16