Quantization publisher

gaianet

gaianet publishes 8 quantizations across 1 models in our index, averaging 5.191 effective bits per weight. Their files differ in size from other publishers' builds of the same nominal quantization on 6 of the pairs we can compare — the same label does not mean the same file.

From the file· summed file bytes
Repositories
1
Quantizations
8
Models covered
1
Avg effective bpw
5.191
across their files

Same model, same quant label, different bytes

largest disagreements first
ModelQuantgaianetvsTheirsDifference
FLUX.1-Fill-devQ8_012.28 GiBYarvixPA11.85 GiB+3.6%
FLUX.1-Fill-devQ8_012.28 GiBsecond-state11.96 GiB+2.6%
FLUX.1-Fill-devQ4_17.14 GiBYarvixPA7.03 GiB+1.6%
FLUX.1-Fill-devQ5_07.83 GiBYarvixPA7.72 GiB+1.4%
FLUX.1-Fill-devQ5_18.52 GiBYarvixPA8.41 GiB+1.3%
FLUX.1-Fill-devQ4_06.45 GiBYarvixPA6.34 GiB+1.8%

A quantization label describes a target, not a recipe. Publishers make different choices about which tensors to keep at higher precision, and some apply an importance matrix while others don't — so two files both honestly labelled the same thing can differ measurably in size and in quality.

Models they publish

ModelQuantizationsSmallest
FLUX.1-Fill-dev83.87 GiB