Downloads · 30 days
72
42% of all-time downloads
Qwen3.6-35B-A3B-oQ4e-mtp
bambocher/Qwen3.6-35B-A3B-oQ4e-mtp
Qwen3.6-35B-A3B-oQ4e-mtp is a machine learning model from bambocher. Use it for the machine learning task on the model card, and read the license before you ship it in a product. It is set up for mlx.
[!NOTE] Updated 2026-10-20: Re-quantized with oMLX v0.7.1rc1. Please re-download if you have an earlier copy.
All-time downloads
170
Public
Parameters
36B
43.2 GB on disk
Likes
0
Public
Repo makeup
Click a slice to open those files.
.safetensors21.6 GB · 100%
Parameter types
How the weights are stored.
U3235.5B · 99%
Model card
From the Hugging Face model README
Qwen3.6-35B-A3B-oQ4e-mtp
[!NOTE] Updated 2026-10-20: Re-quantized with oMLX v0.7.1rc1. Please re-download if you have an earlier copy.
This model was quantized using oQ (oMLX v0.7.1rc1) mixed-precision quantization.
The effective average is 4.809 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 | 33.592B | 93.437% | 17.598 GiB | 4.50 |
| Affine 5-bit, group 128 | 0.575B | 1.599% | 0.351 GiB | 5.25 |
| Affine 6-bit, group 128 | 0.001B | 0.003% | 0.001 GiB | 6.25 |
| Affine 8-bit, group 128 | 0.129B | 0.359% | 0.124 GiB | 8.25 |
| Affine 8-bit, group 64 | 1.177B | 3.274% | 1.165 GiB | 8.50 |
| BF16 | 0.478B | 1.329% | 0.890 GiB | 16.00 |
| Total | 35.952B | 100% | 20.128 GiB | 4.809 |
Storage Breakdown
| Component | Logical weights | Storage |
|---|---|---|
| Language model | 34.661B | 20.216 GB / 18.828 GiB |
| Vision | 0.447B | 0.893 GB / 0.832 GiB |
| MTP head | 0.845B | 0.503 GB / 0.468 GiB |
| Total | 35.952B | 21.613 GB / 20.128 GiB |
- Library
- mlx
- Type
- qwen3_5_moe
- Created
- Aug 18, 2026
- Updated
- Oct 20, 2026