Shiftedx/qwopus3.6-35b-a3b-coder-mxfp4-mlx
Text Generation • 35B • Updated • 286
Qwopus3.6 MLX Quants
Apple Silicon MLX MXFP4/MXFP8 builds for Qwopus3.6 models, including text, vision, and verified native-MTP variants.
Note MXFP4 language-only: smallest local footprint; LM Studio text smoke passed.
Shiftedx/qwopus3.6-35b-a3b-coder-mxfp8-mlx
Text Generation • 35B • Updated • 309 • 2Note MXFP8 language-only: larger, higher-fidelity quant; LM Studio text smoke passed.
Shiftedx/qwopus3.6-35b-a3b-coder-mxfp4-vision-mlx
Image-Text-to-Text • 35B • Updated • 172Note MXFP4 with experimental vision graft; text smoke passed, image smoke not fully reliable.
Shiftedx/qwopus3.6-35b-a3b-coder-mxfp8-vision-mlx
Image-Text-to-Text • 35B • Updated • 364 • 1Note MXFP8 with experimental vision graft; text smoke passed, image smoke partially passed but remains experimental.
Shiftedx/qwopus3.6-27b-fusion-mxfp4-vision-mtplx
Image-Text-to-Text • 27B • Updated • 36 • 1Note 27B Fusion MXFP4 vision build with verified native BF16 MTP; D3 measured 1.79x AR locally.
Shiftedx/qwopus3.6-27b-fusion-mxfp4-vision-mlx
Image-Text-to-Text • 27B • Updated • 39 • 1Note 27B Fusion MXFP4 vision-enabled MLX build; direct image and OCR smokes passed.
Shiftedx/qwopus3.6-27b-fusion-mxfp4-text-mlx
Text Generation • 27B • Updated • 30Note 27B Fusion MXFP4 language-only MLX build; smallest footprint and fastest tested body.
Shiftedx/qwopus3.6-27b-fusion-mxfp8-vision-mtplx
Image-Text-to-Text • 27B • Updated • 46 • 1Note 27B Fusion MXFP8 vision build with verified native BF16 MTP; D3 measured 1.47x AR locally.
Shiftedx/qwopus3.6-27b-fusion-mxfp8-vision-mlx
Image-Text-to-Text • 27B • Updated • 35Note 27B Fusion MXFP8 vision-enabled MLX build; direct image and OCR smokes passed.
Shiftedx/qwopus3.6-27b-fusion-mxfp8-text-mlx
Text Generation • 27B • Updated • 56Note 27B Fusion MXFP8 language-only MLX build; higher-precision standard runtime.