ComfyUI OOMs after 20 generations: MiniMaxH3 suspected memory leak
reicaden · reddit · 2026-08-17
User reports a consistent CUDA out of memory error in ComfyUI when using the MiniMaxH3 model after 20-30 successful generations. Logs indicate high memory usage during the loading of MiniMaxH3VideoVAE and MiniMaxH3TEModel. Standard unloading methods fail, requiring a full restart to restore functionality.
More from Infra
- antirez's DwarfStar: adaptive VRAM/RAM expert placement runs DeepSeek V4 at 45 t/s — antirez · 2026-08-17
- DeepSeek v4 PRO Q2 hits 45 t/s with VRAM/RAM split on DGX Station — antirez · 2026-08-17
- Model routing should be optimized at the harness layer, not the gateway — agihouse_org · 2026-08-17
- Qwen3.8-27B Benchmarks: RPC inference tests across AMD and Nvidia GPUs — tabletuser_blogspot · 2026-08-17
- a16z: How Neoclouds Transformed from Crypto Mining to AI Compute Powerhouses — rohanpaul_ai · 2026-08-17
- AI Spending Gap 625x: Top 1% Spend $7,500/Employee/Month vs Median $12 — rohanpaul_ai · 2026-08-17