| File | ltx-2.5-22b-dev-transformer-bf16.safetensors |
| Model | LTX-2.5 (22B) (video model, 22B) |
| Precision | 16-bit |
| Size | 42.0 GB |
| VRAM to fit | about 46.8 GB |
| Folder | ComfyUI/models/diffusion_models/ |
| Loader node | Load Diffusion Model (UNETLoader) |
| Source | Lightricks/LTX-2.5 on Hugging Face |
Download ltx-2.5-22b-dev-transformer-bf16.safetensors (42.0 GB)
Where to put it
Put the file in ComfyUI/models/diffusion_models/ (portable version: ComfyUI_windows_portable/ComfyUI/models/diffusion_models/). Older guides say models/unet: that folder still works. Press R in ComfyUI or restart it, then pick the file in Load Diffusion Model (UNETLoader).
If a workflow says “Value not in list” for that node, the workflow expects a file with another name. Click the node and choose this one.
Which graphics cards it fits
| VRAM | This file | Best file for that card (RTX 40/50) | |
|---|---|---|---|
| 6 GB | Not practical | Short by 40.8 GB | LTX-2.5-Distilled-Q2_K.gguf (8.8 GB) |
| 8 GB | Not practical | Short by 38.8 GB | LTX-2.5-Distilled-Q4_K_M.gguf (15.1 GB) |
| 10 GB | Not practical | Short by 36.8 GB | LTX-2.5-Distilled-Q4_K_M.gguf (15.1 GB) |
| 12 GB | Not practical | Short by 34.8 GB | LTX-2.5-Distilled-Q2_K.gguf (8.8 GB) |
| 16 GB | Not practical | Short by 30.8 GB | LTX-2.5-Distilled-Q2_K.gguf (8.8 GB) |
| 20 GB | Not practical | Short by 26.8 GB | LTX-2.5-Distilled-Q4_K_M.gguf (15.1 GB) |
| 24 GB | Offload only | Short by 22.8 GB: runs with part of it in system RAM, slower | LTX-2.5-Distilled-Q6_K.gguf (18.7 GB) |
| 32 GB | Offload only | Short by 14.8 GB: runs with part of it in system RAM, slower | LTX-2.5-Distilled-Q8_0.gguf (23.6 GB) |
| 48 GB | Runs well | Fits in VRAM, 1.2 GB to spare | this file |
Need = file × 1 + 4 GB for the picture and 0.8 GB for Windows and the browser — the same rule as the rest of the site.
What else you need
This file alone does not make a picture. With it you load:
| Role | File | Size | Folder |
|---|---|---|---|
| Model | ltx-2.5-22b-dev-transformer-bf16.safetensors | 42.0 GB | models/diffusion_models |
| Text encoder | gemma4-12b-with-proj-ltx-2.5-comfy-int8-convrot.safetensors | 15.4 GB | models/text_encoders |
| VAE | ltx-2.5-video-vae-bf16.safetensors | 1.5 GB | models/vae |
| Also needed | ltx-2.5-audio-vae-bf16.safetensors | 0.4 GB | models/vae |
59.2 GB in total. Smaller text-encoder versions and other options are on the LTX-2.5 page.
Other files of LTX-2.5
| File | Precision | Size | VRAM to fit |
|---|---|---|---|
| ltx-2.5-22b-dev-transformer-bf16.safetensors | 16-bit | 42.0 GB | 46.8 GB |
| LTX-2.5-Distilled-Q8_0.gguf | Q8_0 | 23.6 GB | 28.4 GB |
| LTX-2.5-Distilled-Q6_K.gguf | Q6_K | 18.7 GB | 23.5 GB |
| LTX-2.5-Distilled-Q5_K_M.gguf | Q5_K_M | 16.8 GB | 21.6 GB |
| LTX-2.5-Distilled-Q4_K_M.gguf | Q4_K_M | 15.1 GB | 19.9 GB |
| LTX-2.5-Distilled-Q3_K_M.gguf | Q3_K_M | 11.5 GB | 16.3 GB |
| LTX-2.5-Distilled-Q2_K.gguf | Q2_K | 8.8 GB | 13.6 GB |
Same model, different compression. 8-bit files (FP8, INT8, Q8_0) look almost the same as 16-bit; below Q5 the loss starts to show.
Want the answer for your exact card? Check your GPU. Every file of every model: ComfyUI model files.
Size and source read from Hugging Face (2026-10-01).