Quantization publisher

wikeeyang

wikeeyang publishes 18 quantizations across 4 models in our index, averaging 6.544 effective bits per weight. Their files differ in size from other publishers' builds of the same nominal quantization on 24 of the pairs we can compare — the same label does not mean the same file.

From the file· summed file bytes
Repositories
5
Quantizations
18
Models covered
4
Avg effective bpw
6.544
across their files

Same model, same quant label, different bytes

largest disagreements first
ModelQuantwikeeyangvsTheirsDifference
FLUX.2-klein-9BQ8_09.29 GiBponpoke8.11 GiB+14.6%
FLUX.2-klein-9BQ8_09.29 GiBponpoke8.11 GiB+14.6%
FLUX.2-klein-9BQ6_K7.33 GiBponpoke6.26 GiB+16.9%
FLUX.2-klein-9BQ6_K7.33 GiBponpoke6.26 GiB+16.9%
Krea-2-TurboQ4_K_S7.65 GiBrealrebelai6.76 GiB+13.1%
Krea-2-TurboQ4_K_S7.65 GiBvantagewithai6.97 GiB+9.7%
Krea-2-TurboQ5_K_S8.86 GiBrealrebelai8.21 GiB+7.9%
Krea-2-TurboQ5_K_S8.86 GiBrealrebelai8.26 GiB+7.2%
Krea-2-TurboQ5_K_S8.86 GiBvantagewithai8.26 GiB+7.2%
Wan2.2-T2V-A14BQ4_18.99 GiBQuantStack8.62 GiB+4.2%
Wan2.2-T2V-A14BQ4_18.99 GiBQuantStack8.62 GiB+4.2%
Krea-2-TurboQ6_K10.15 GiBrealrebelai9.80 GiB+3.6%
Krea-2-TurboQ6_K10.15 GiBrealrebelai9.86 GiB+3.0%
Krea-2-TurboQ6_K10.15 GiBvantagewithai9.86 GiB+2.9%
Krea-2-TurboQ8_012.97 GiBrealrebelai12.69 GiB+2.2%
Krea-2-TurboQ8_012.97 GiBrealrebelai12.76 GiB+1.6%
Krea-2-TurboQ8_012.97 GiBvantagewithai12.76 GiB+1.6%
Qwen-Image-2512Q4_K_M12.17 GiBunsloth12.34 GiB-1.4%
Wan2.2-T2V-A14BQ8_014.35 GiBQuantStack14.35 GiB-0.0%
Wan2.2-T2V-A14BQ8_014.35 GiBQuantStack14.35 GiB-0.0%
FLUX.2-klein-9BQ6_K7.33 GiBunsloth7.33 GiB0.0%
FLUX.2-klein-9BQ8_09.29 GiBunsloth9.29 GiB0.0%
FLUX.2-klein-9BQ8_09.29 GiBunsloth9.29 GiB0.0%
FLUX.2-klein-9BQ6_K7.33 GiBunsloth7.33 GiB0.0%

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.2-klein-9B85.32 GiB
Krea-2-Turbo67.34 GiB
Wan2.2-T2V-A14B28.99 GiB
Qwen-Image-2512212.17 GiB