ComfyUI Node
LC Model Optimizer Video (pipe) ⚡
This node analyzes your machine and loads the model, text encoder, video and audio VAE and latent upscaler that suit YOUR card, with the speed-ups that actually work on it (Sage, Comfy Kitchen, etc.). Run the report with the button below, or from the LC123 settings. Missing a file? Pick a ⬇ entry and it downloads on the first run. Finetunes work too. Pick your own file and it still gets the speed-ups, plus a ⚠️ if the text encoder or VAE doesn't match. Any other model: set the base model to Custom and pick your own files (and the text encoder type). The speed-ups still apply; recommendations, estimates and checks show as unsupported. Current available models: MiniMax H3, LTX 2.5, LTX 2.3, Custom.
LC Model Optimizer Video (pipe) ⚡
- pipe
- model
- clip
- vae
- audio_vae
- latent_upscaler
- summary
◄base_modelMiniMax H3 FL2VA►
◄goalOptimal►
◄model_file★ Recommended►
◄text_encoder★ Recommended►
◄vae★ Recommended►
◄audio_vae★ Recommended►
◄latent_upscaler★ Recommended►
◄lora★ Recommended►
◄speed_upsAuto►
◄attentionauto►
◄fp16_accumulationauto►
◄step_cacheauto►
◄torch_compileoff►
◄clip_typeauto►
CategoryLC123/optimizer
Inputs (14)
| Name | Type | Default | Description |
|---|---|---|---|
| base_model | COMBO | MiniMax H3 FL2VA | The model family. Picks the recommended files and the checks. |
| goal | COMBO | Optimal | Quality = closest to bf16. Optimal = fast and near identical. Fast = fastest that still looks right. |
| model_file | COMBO | ★ Recommended | 52 options: ★ Recommended, From checkpoint, None, ⬇ Download: ltx-2.3-22b-dev.safetensors, ⬇ Download: ltx-2.3-22b-dev-fp8.safetensors, ⬇ Download: ltx-2.3-22b-dev-nvfp4.safetensors, +46 |
| text_encoder | COMBO | ★ Recommended | 17 options: ★ Recommended, From checkpoint, None, ⬇ Download: gemma_3_12B_it.safetensors, ⬇ Download: gemma_3_12B_it_fp8_scaled.safetensors, ⬇ Download: gemma_3_12B_it_fp4_mixed.safetensors, +11 |
| vae | COMBO | ★ Recommended | 6 options: ★ Recommended, From checkpoint, None, ⬇ Download: ltx-2.5-video-vae-bf16.safetensors, ⬇ Download: minimax_h3_video_vae_int8_convrot.safetensors, ⬇ Download: minimax_h3_video_vae_fp16.safetensors |
| audio_vae | COMBO | ★ Recommended | 5 options: ★ Recommended, From checkpoint, None, ⬇ Download: ltx-2.5-audio-vae-bf16.safetensors, ⬇ Download: minimax_h3_audio_vae_fp32.safetensors |
| latent_upscaler | COMBO | ★ Recommended | 5 options: ★ Recommended, From checkpoint, None, ⬇ Download: ltx-2.3-spatial-upscaler-x2-1.1.safetensors, ⬇ Download: ltx-2.5-spatial-upscaler-x2.safetensors |
| lora | COMBO | ★ Recommended | 4 options: ★ Recommended, From checkpoint, None, ⬇ Download: ltx_2.3_22b_distilled_1.1_lora_dynamic_fro09_avg_rank_111_bf16.safetensors |
| speed_ups | COMBO | Auto | Auto = from the System & Model Optimization Report. Manual = choose each one below. |
| attention | COMBO | auto | Attention backend. auto = the fastest one that passed the report's test. |
| fp16_accumulation | COMBO | auto | Faster fp16 math with a small quality cost. auto = Fast goal only. |
| step_cache | COMBO | auto | EasyCache: skips steps that barely change. auto = Fast goal, 20+ steps. |
| torch_compile | COMBO | off | Compiles the model: the first run is slow, later runs are faster. Needs Triton. |
| clip_type | COMBO | auto | Custom only: how the text encoder is loaded (CLIPLoader's type). auto = from the encoder's shape when it matches one known base model. Ignored for the other base models and for a checkpoint's own encoder. |
Outputs (7)
| Name | Type | Description |
|---|---|---|
| pipe | LC_PIPE | — |
| model | MODEL | — |
| clip | CLIP | — |
| vae | VAE | — |
| audio_vae | VAE | — |
| latent_upscaler | LATENT_UPSCALE_MODEL | — |
| summary | STRING | — |