Brooooooklyn/Qwen3.6-27B-UD-Q2_K_XL-mlx

Qwen3.6-27B — UD-Q2_K_XL (mlx-node)

2-bit base mixed-precision quantization of Qwen/Qwen3.6-27B for Apple Silicon, using the Unsloth Dynamic quantization strategy via mlx-node.

	Original (BF16)	This Model
Size	~51 GB	15 GB
Format	SafeTensors (sharded)	SafeTensors (sharded)
Precision	BF16 uniform	Mixed 2-bit + BF16

All Variants

Repo	GGUF Equivalent	Size	Decode (tok/s)
Brooooooklyn/Qwen3.6-27B-UD-Q2_K_XL-mlx (this model)	UD-Q2_K_XL	15 GB	20.0
Brooooooklyn/Qwen3.6-27B-UD-Q3_K_XL-mlx	UD-Q3_K_XL	18 GB	16.2
Brooooooklyn/Qwen3.6-27B-UD-NVFP4_K_XL-mlx	—	21 GB	14.9
Brooooooklyn/Qwen3.6-27B-UD-MXFP4_K_XL-mlx	—	21 GB	15.9
Brooooooklyn/Qwen3.6-27B-UD-Q4_K_XL-mlx	UD-Q4_K_XL	21 GB	15.3
Brooooooklyn/Qwen3.6-27B-UD-Q5_K_XL-mlx	UD-Q5_K_XL	25 GB	13.4
Brooooooklyn/Qwen3.6-27B-UD-Q6_K_XL-mlx	UD-Q6_K_XL	27 GB	12.4
Brooooooklyn/Qwen3.6-27B-UD-MXFP8_K_XL-mlx	—	29 GB	10.5
Brooooooklyn/Qwen3.6-27B-UD-Q8_K_XL-mlx	UD-Q8_K_XL	30 GB	9.9

Benchmarked on Apple M3 Max 128GB via examples/lm.ts (best decode tok/s across turns 2–4, steady-state).

Performance

Steady-state decode: 20.0 tok/s on Apple M3 Max 128GB (best of turns 2–4, examples/lm.ts capitals chat with reasoningEffort: 'low'). Decode is memory-bandwidth bound on Apple Silicon — fewer bytes per token directly translates to higher throughput.

Per-Tensor Bit Assignments (N=2)

Weight	Bits	Rationale
embed_tokens	4-bit	KLD ~0.15 — very low sensitivity
lm_head	5-bit	KLD ~0.05 — safest tensor
self_attn.q/k/v_proj	4-bit + AWQ	KLD ~1.5–2.9, AWQ via layernorm
linear_attn.in_proj_qkv/z	4-bit + AWQ	KLD ~2.9, AWQ via layernorm
self_attn.o_proj	bf16	NOT AWQ-correctable
linear_attn.out_proj	bf16	KLD ~6.0 — worst tensor
down_proj	3-bit	"Slightly more sensitive"
gate_proj, up_proj	2-bit	base bits
GDN params (A_log, etc)	bf16	State-space dynamics

Quantization Strategy

Based on Unsloth Dynamic 2.0 per-tensor KLD analysis. Sensitive layers get higher bits with AWQ correction, while the bulk of FFN weights are aggressively quantized. imatrix AWQ pre-scaling amplifies important weight channels and fuses inverse scales into preceding layer norms (zero inference overhead).

AWQ-correctable projections (q/k/v, in_proj_qkv/z) are quantized at 4-bit via input_layernorm. Non-AWQ-correctable projections (o_proj, out_proj) are kept at bf16 — their inputs come from attention/GDN computation, not from a norm layer.

Architecture

Parameter	Value
Total parameters	27.4B (dense — all active)
Hidden size	5,120
Layers	64 (48 linear + 16 full attention)
Attention heads	24 (4 KV heads, GQA 6:1)
Head dimension	256
Intermediate size	17,408
Vocab size	248,320
Max context	262,144 tokens

Usage

import { loadSession } from '@mlx-node/lm';

const session = await loadSession('./Qwen3.6-27B-UD-Q2_K_XL-mlx');

for await (const event of session.sendStream('Explain the hybrid attention mechanism in Qwen3.6.', {
  config: { maxNewTokens: 2048, temperature: 0.6, reasoningEffort: 'low' },
})) {
  if (!event.done) process.stdout.write(event.text);
}

How It Was Made

mlx convert \
  -i Qwen3.6-27B \
  -o Qwen3.6-27B-UD-Q2_K_XL-mlx \
  -q --q-bits 2 --q-recipe unsloth \
  --imatrix-path imatrix_unsloth.gguf

Acknowledgments

• Unsloth — Quantization strategy based on their per-layer KLD benchmarks and Dynamic 2.0 methodology
• Qwen Team — For the Qwen3.6 model family
• Apple MLX — For the Metal-accelerated ML framework

License

Apache 2.0 (inherited from base model).