GingerLabsPlatform/GingerLabs-Wan2.2-T2V-A14B

GingerLabs Wan 2.2 T2V A14B

A GingerLabs-tuned derivative of Wan-AI/Wan2.2-T2V-A14B, packaged as quantized HighNoise and LowNoise diffusion-model pairs for ComfyUI.

This is a quantized-weights-only release. It does not include the VAE, text encoder, or other pipeline assets required for generation.

Content warning: This model may generate mature, explicit, or NSFW material, including unintended outputs. Use it only in lawful, consensual, age-appropriate contexts.

Download a matching pair

Wan 2.2 T2V A14B uses separate HighNoise and LowNoise models. Download both files for one quantization tier and keep the tiers matched—for example, High_Q5_K_M with Low_Q5_K_M.

Variant	Format	Size per file	Pair size	ComfyUI loader
FP8 E4M3FN scaled	Safetensors	13.97 GiB	27.94 GiB	Native scaled-FP8 loader or WanVideoWrapper
Q8_0	GGUF	14.35 GiB	28.69 GiB	ComfyUI-GGUF
Q6_K	GGUF	11.18 GiB	22.36 GiB	ComfyUI-GGUF
Q5_K_M	GGUF	10.05 GiB	20.10 GiB	ComfyUI-GGUF
Q4_K_M	GGUF	8.99 GiB	17.97 GiB	ComfyUI-GGUF

Sizes use GiB (2^30 bytes).

Repository layout

HighNoise/
  GingerLabs_Wan2.2_T2V_A14B_High_FP8_E4M3FN_SCALED.safetensors
  GingerLabs_Wan2.2_T2V_A14B_High_Q8_0.gguf
  GingerLabs_Wan2.2_T2V_A14B_High_Q6_K.gguf
  GingerLabs_Wan2.2_T2V_A14B_High_Q5_K_M.gguf
  GingerLabs_Wan2.2_T2V_A14B_High_Q4_K_M.gguf
LowNoise/
  GingerLabs_Wan2.2_T2V_A14B_Low_FP8_E4M3FN_SCALED.safetensors
  GingerLabs_Wan2.2_T2V_A14B_Low_Q8_0.gguf
  GingerLabs_Wan2.2_T2V_A14B_Low_Q6_K.gguf
  GingerLabs_Wan2.2_T2V_A14B_Low_Q5_K_M.gguf
  GingerLabs_Wan2.2_T2V_A14B_Low_Q4_K_M.gguf

Download example

hf download GingerLabsPlatform/GingerLabs-Wan2.2-T2V-A14B \
  HighNoise/GingerLabs_Wan2.2_T2V_A14B_High_Q5_K_M.gguf \
  LowNoise/GingerLabs_Wan2.2_T2V_A14B_Low_Q5_K_M.gguf \
  --local-dir GingerLabs-Wan2.2-T2V-A14B

Replace Q5_K_M with the same desired variant in both filenames.

ComfyUI usage

GGUF

1. Install ComfyUI-GGUF.
2. Put the matching HighNoise and LowNoise .gguf files in ComfyUI/models/diffusion_models/ (ComfyUI/models/unet/ on older installations).
3. Load each file with Unet Loader (GGUF) in a Wan 2.2 T2V workflow.

Scaled FP8

1. Put the matching .safetensors files in ComfyUI/models/diffusion_models/.
2. Use a scaled-FP8-aware loader. With ComfyUI-WanVideoWrapper, select fp8_e4m3fn_scaled.
3. Do not treat these files as raw, unscaled E4M3FN weights.

The FP8 filenames omit _KJ, but their internal scale tensors and marker follow the Kijai/Tencent-compatible scaled-FP8 convention.

Quantization details

The FP8 files store 400 selected attention and feed-forward matrices as E4M3FN with per-matrix FP32 scale weights. The other 695 source tensors are preserved as FP32.

The GGUF files were converted with ComfyUI-GGUF and its patched llama.cpp toolchain. The five-dimensional patch_embedding.weight tensor was restored after quantization for Wan/ComfyUI compatibility.

Validation

All ten files passed structural validation and 12 CUDA load/unload checks:

• FP8 through native ComfyUI and ComfyUI-WanVideoWrapper
• GGUF through ComfyUI-GGUF
• 1,496 tensors in each FP8 file
• 1,095 unique tensors in each GGUF file
• Restored five-dimensional GGUF patch-embedding tensor
• No missing or duplicate model keys

Validation was load-level only. No full video generation, visual-quality comparison, or performance benchmark was performed.

See manifest.json for repository-relative provenance and validation records, and checksums.sha256 for weight hashes.

Limitations

Quantization can change output quality, prompt adherence, motion, and numerical behavior. Lower-bit variants generally reduce memory requirements at a greater potential quality cost. Results also depend on the workflow, sampler, resolution, frame count, and supporting model assets.

Attribution and terms

This release is based on Wan-AI/Wan2.2-T2V-A14B, whose repository identifies the base model as Apache-2.0 licensed. Quantization and compatibility tooling are credited in NOTICE.md.

No new license claim is made by this model card. Use and redistribution remain subject to the terms applicable to the base model and all tuning or merge components.