Download
5 variants available
mxfp8 SafeTensor
Chimera_v1_mxfp8.safetensors
Block-scaled 8-bit, near-FP16 quality • 12.91 GB
Verified: a month ago
SafeTensor
mxfp8
Chimera_v1_mxfp8.safetensors
Block-scaled 8-bit, near-FP16 quality
Verified: a month ago

1.2K0 1 2 3 4 5 6 7 8 9.0 1 2 3 4 5 6 7 8 9K
4.9K0 1 2 3 4 5 6 7 8 9.0 1 2 3 4 5 6 7 8 9K

License:
This started as a fun experiment with one simple rule: "Nobody touches the DiT booty." 🏴☠️
The pirate insisted.
Project Chimera is built on a custom Turbo/Raw hybrid DiT, not a Turbo LoRA or simple weighted merge. The primary innovation is extensive modification of the TextFusion (CLIP) components, while the underlying hybrid DiT architecture itself remains unchanged.
⚠️ Requirements for w4a8
Please make sure you're using ComfyUI v0.31.0 or newer. Older builds may not include the required W4A8 support.
Low-VRAM users: I also made a W4A8 quantized Qwen 3 VL 4B text encoder, only 2.86 GB. You can find it here: https://civitai.red/models/2843533/qwen-3-vl-4b-w4a8?modelVersionId=3210230

