ltx-2.3-22b-distilled_transformer_only_fp8_scaled.safetensors
LTX 2.3 Distilled FP8 Scaled (Kijai)
FP8 scaled distilled by Kijai. Alternative FP8 quantization method.
Released 2026-03 · Source: Kijai/LTX2.3_comfy (HuggingFace) — v1.0 weight-scaled FP8 distilled. Superseded by the v1.1 file on 2026-04-13.
Download ltx-2.3-22b-distilled_transformer_only_fp8_scaled.safetensors
Direct HuggingFace download. ~25 GB · Free.
No 16GB GPU? Try ltx-2.3-22b-distilled_transformer_only_fp8_scaled.safetensors online — free generation included
Skip the ~25 GB download and ComfyUI setup. Generate a 6-second video using this exact model in your browser, ~30 seconds.
Technical details
This is the v1.0 distilled checkpoint quantized to FP8 with weight-side scaling (as opposed to fp8_input_scaled which scales activations). Same v1.0 weights, different quantization axis — and a different file from fp8_input_scaled_v3, despite the similar size.
Weight-scaled FP8 is slightly faster at inference because the scale factors fold into the matmul kernel itself, no extra activation pass needed. Memory footprint is identical. Numerical behavior diverges slightly from input_scaled in attention layers; neither has a consistent quality advantage on standard prompts.
'transformer_only' — pair with taeltx2_3.safetensors VAE and a Gemma 3 12B text encoder. Distilled inference settings: 8 steps, CFG=1.
When to choose ltx-2.3-22b-distilled_transformer_only_fp8_scaled.safetensors
Use this when you specifically need weight-scaled FP8 of the v1.0 distilled model — for reproducing a workflow JSON that hardcodes this filename, or running an experiment comparing scaling axes.
For day-to-day generation, ltx-2.3-22b-distilled-1.1_transformer_only_fp8_scaled.safetensors is strictly better: same scaling style, newer (v1.1) weights, better fast-motion handling. Same VRAM, same speed.
On RTX 30-series, switch to the MXFP8 distilled v1.1 file — Ampere can't do FP8 matmul natively.
Will this run on my GPU?
Minimum: 16GB VRAM. Headroom up to: 24GB.
⚠ FP8 scaled matmul requires RTX 40-series or newer (Ada Lovelace architecture). RTX 30xx cannot run this format — use the MXFP8 block-32 or BF16 variant instead.
Recommendation: Alternative FP8 quantization. Use v1.1 FP8 for latest quality.
How to use ltx-2.3-22b-distilled_transformer_only_fp8_scaled.safetensors
- Download the file from HuggingFace.
- Place it in ComfyUI/models/checkpoints/ inside your ComfyUI directory.
- Restart ComfyUI (or refresh the model list from the menu).
- Load a compatible workflow — see below.
Compatible official workflows:
- LTX-2.3_T2V_I2V_Single_Stage_Distilled_Full.json— T2V / I2V Single Stage Distilled
- LTX-2.3_T2V_I2V_Two_Stage_Distilled.json— T2V / I2V Two Stage Distilled
- LTX-2.3_ICLoRA_Union_Control_Distilled.json— ICLoRA Union Control Distilled
- LTX-2.3_ICLoRA_Motion_Track_Distilled.json— ICLoRA Motion Track Distilled
- LTX-2.3_ICLoRA_HDR_Distilled.json— ICLoRA HDR Distilled
Don't want to run this locally? Try ltx-2.3-22b-distilled_transformer_only_fp8_scaled.safetensors online with a free generation — no GPU, no install, ~30 seconds per clip.
ComfyUI says it can't find ltx-2.3-22b-distilled_transformer_only_fp8_scaled.safetensors?
Some published workflow JSONs reference this file under a custom subdirectory. If ComfyUI shows a "cannot find model" error and your workflow references one of these path-prefixed variants:
- ltx\ltx-2.3-22b-distilled_transformer_only_fp8_scaled.safetensors
- diffusion_models/ltx-2.3-22b-distilled_transformer_only_fp8_scaled.safetensors
The prefix before the slash or backslash is a subdirectory the workflow author used. The actual file is the same ltx-2.3-22b-distilled_transformer_only_fp8_scaled.safetensors — you have two fixes:
- Create the matching subdirectory inside ComfyUI/models/checkpoints/ and place the file there. Example: if the workflow references ltx\ltx-2.3-22b-distilled_transformer_only_fp8_scaled.safetensors, create the corresponding subfolder under ComfyUI/models/checkpoints/ and put ltx-2.3-22b-distilled_transformer_only_fp8_scaled.safetensors inside it.
- Or open the workflow JSON in a text editor and replace the prefixed string with just ltx-2.3-22b-distilled_transformer_only_fp8_scaled.safetensors. ComfyUI then resolves it directly from ComfyUI/models/checkpoints/.
On Windows the separator is \, on macOS/Linux it is / — they refer to the same nested folder regardless of platform.
Common issues
I have both this file and fp8_input_scaled_v3 — which does ComfyUI pick?▼
ComfyUI lists both in the loader dropdown. The workflow JSON's hardcoded string determines which one loads. Fix: Open the workflow JSON and look for the .safetensors filename — that's the one that'll load. Or in ComfyUI, manually select the file you want in the loader node.
Generation is the same speed as v1.1 fp8_scaled but output is worse▼
This is v1.0 distilled. v1.1 trained on more data with refined distillation hyperparameters. Fix: Switch to ltx-2.3-22b-distilled-1.1_transformer_only_fp8_scaled.safetensors. The quality gap is small but consistent.
ComfyUI doesn't see the file after I downloaded it▼
Make sure the file is in ComfyUI/models/checkpoints/ (not a subfolder). Restart ComfyUI fully — the menu refresh sometimes misses new files. Filename must match exactly: ltx-2.3-22b-distilled_transformer_only_fp8_scaled.safetensors.
I get a CUDA error mentioning fp8 / scaled / matmul▼
FP8 scaled matmuls require an RTX 40-series GPU or newer (Ada Lovelace architecture). RTX 30-series and older cannot run FP8 weights at native precision. Use the BF16 variant instead, or the MXFP8 block-32 alternative.
CUDA out of memory error when loading the model▼
ltx-2.3-22b-distilled_transformer_only_fp8_scaled.safetensors needs ~16GB VRAM minimum. If you're hitting OOM: • Enable Sequential Offloading in ComfyUI settings • Lower the resolution (768×512 instead of 1280×704) — both dimensions must be divisible by 32 • Reduce frame count (65 frames instead of 161) — must be 8n+1 • Use a smaller variant — see Related models below.
Get notified when LTX 2.3 Distilled FP8 Scaled (Kijai) updates
Occasional updates on what's new in LTX 2.3 — new FP8 quants, LoRAs, IC-LoRA releases — with our hands-on verdict on whether they're worth re-downloading. No fixed cadence.
No spam. Sent occasionally when there's real news. Unsubscribe in one click.
Related models
ltx-2.3-22b-distilled-1.1_transformer_only_fp8_scaled.safetensors
ltx-2.3-22b-distilled-1.1_transformer_only_mxfp8_block32.safetensors
ltx-2.3-22b-distilled_transformer_only_fp8_input_scaled_v3.safetensors
ltx-2.3-22b-dev_transformer_only_fp8_input_scaled.safetensors
ltx-2.3-22b-dev_transformer_only_fp8_scaled.safetensors
ltx-2.3-22b-dev_transformer_only_mxfp8_block32.safetensors