# Provenance — majentik/Qwen3.6-27B-MLX-2bit

- Upstream repo: `Qwen/Qwen3.6-27B`
- Upstream revision: `6a9e13bd6fc8f0983b9b99948120bc37f49c13e9`
- Quantization: bits=2, mode=affine, group_size=64
- Command: `python -m pipelines.mlx_direct_quantize --model qwen3.6-27b --base-dir /private/tmp/mlx-direct-release/qwen3.6-27b/base --out-dir /tmp/mlx-direct-release/qwen3.6-27b/2bit --bits 2 --mode affine --group-size 64`
- Input bytes (BF16 base): 55586111229
- Output bytes (this variant): 9487101205
- Toolchain: mlx 0.31.2, huggingface_hub 0.36.2
- Generated: 2026-07-07T07:30:40.513390+00:00

No local inference smoke was performed on these artifacts; runtime support pending upstream mlx-lm support for the `qwen3_5` vision-language architecture.