For privacy reasons a browser tells us at most
"≥ 8 GB RAM, 8 cores" - same reading whether you have 8 GB
or 128 GB. It has no idea how much RAM is free right now, which
apps are open, or whether you have a GPU.
The Abliteration app is integrated with your machine
It reads your exact RAM, GPU model and VRAM,
free memory right now, and picks the sharpest quant that
still fits. Every model page lights up precisely for your rig.
And you can chat with any model, right now
The app is a full local runtime - no API keys, no subscription,
everything runs on your machine. Click any model on this site and
start a conversation in seconds.
This model was quantized using oQ (oMLX v0.8.0) mixed-precision quantization.
The effective average is 23.208 bits/weight excluding the N-gram embedding table, or 12.645 bits/weight for the complete checkpoint.
Quantization and Bit Distribution
Storage format
Logical weights
Share of total
Tensor storage
Effective bits/weight
Affine 4-bit, group 64
2.601B
3.528%
1.363 GiB
4.50
Affine 6-bit, group 64
0.883B
1.198%
0.668 GiB
6.50
Affine 8-bit, group 128
0.241B
0.327%
0.231 GiB
8.25
Affine 8-bit, group 64
3.763B
5.105%
3.723 GiB
8.50
FP8 (E4M3)
51.200B
69.456%
47.684 GiB
8.00
BF16
0.610B
0.828%
1.137 GiB
16.00
U32
14.418B
19.559%
53.711 GiB
32.00
I64
0.000B
0.000%
0.000 GiB
64.00
Total
73.717B
100%
108.518 GiB
12.645
Storage Breakdown
Component
Logical weights
Storage
Language model
19.460B
62.951 GB / 58.628 GiB
N-gram embedding
51.200B
51.200 GB / 47.684 GiB
Vision
0.449B
0.898 GB / 0.836 GiB
MTP head
2.607B
1.471 GB / 1.370 GiB
Total
73.717B
116.520 GB / 108.518 GiB
README history
1 version
The author's README evolved over time. Click a version to see its content at that point.
2026-11-20Upload Qwen3.8-Flash-Next-Uncensored-oQ3.5h-mtp-plefp8 via oMLX5200eb61.3 KB
Loading...
Catalog is the map. Apps are the tools.
Run models on your own machine, not in the cloud.
Every model page has an "Open in Abliteration" button that hands the model directly to the first-party desktop client, at the quantization your rig can actually run. No API keys, no subscription, no prompt leakage.