Wan 2.1 T2V 1.3B — BonfyreFPQ v12 Native
- Format: .fpq v12 (rANS entropy coded E8 + 6-bit tiles + FP16 scales)
- Base model: Wan-AI/Wan2.1-T2V-1.3B (1.42B params, 825 tensors)
- Size: 1.3 GB (vs 5.3 GB FP32 = 4.2× compression)
- Bits per weight: 7.53 bpw
- Quality: avg cosine 0.9999, worst 0.9996
- Video generation verified: decoded → diffusers WanPipeline → 480×832 video, 30 steps
Decode + Generate
./bonfyre-fpqx decode diffusion_pytorch_model.fpq # outputs BF16 safetensors
# Then load into diffusers WanPipeline (key remapping needed — see bonfyre docs)
- Downloads last month
- -
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support
Model tree for NICKO/Wan2.1-T2V-1.3B-BonfyreFPQ-v12
Base model
Wan-AI/Wan2.1-T2V-1.3B