RunPod Test: Generating 30s Video with MiniMax H3 on A100 Costs Just $0.72
shadowtheimpure · reddit · 2026-08-04
A developer tested the open-source video model MiniMax H3 (FL2VA Pruned Int8) on a rented A100 (80GB) via RunPod to measure local generation costs.
- Parameters: 960x544 resolution, text-to-video (30 seconds long).
- Time & Cost: Averaged 92 seconds per iteration, totaling 34m12s. Billable time from pod launch to completion was 46m40s, resulting in a total session cost of just $0.723.
- Prompting: Used ChatGPT to generate a complex, multi-shot action prompt based on official guidelines, successfully driving cinematic outputs.
More from Infra
- Dynamic Model Router Matches Inference Levels Based on Context and KV Cache — testingcatalog · 2026-08-05
- Open-Source 'Personal AI Computer': Build a Local AI Data Center with RTX 5090s — dee_hw · 2026-08-05
- AI Compute Demand to Quadruple Data Center Electricity by 2030 — bittingthembits · 2026-08-05
- DeepSeek V4 Flash Local Deployment Hits 16k Output Limit — El_90 · 2026-08-05
- China's Potential InP Wafer Export Ban Threatens Global AI Infrastructure — teortaxesTex · 2026-08-05
- 3-Hour Deep Dive with vLLM Core Dev: Open Source Infra and Model Co-design — vista8 · 2026-08-05