Qwen3.6-35B-A3B (35B params, 3B active per token, Qwen3.5-MoE family) quantized to Q6_0_ROCMFPX (~6.5 BPW) — the quality tier of the ROCmFPX fork family, which we serve and benchmark on our lab runtime (full source: pugant/strix-nebulosa; upstream: charlie12345/ROCmFPX). This is the highest-quality ROCmFPX preset for this model, targeting coding-agent workloads where quality matters more than raw decode speed. MTP (multi-token prediction) layers are included: serve with --spec-type draft-mtp --spec-draft-n-max 3 to enable speculative decoding (up to 67 tok/s on deterministic content, +37% over plain tg128).

⚠️ This GGUF is for the ROCmFPX fork of llama.cpp. It will not load in stock llama.cpp.

⚠️ Critical warnings — read before use

Q6_0_ROCMFPX is a fork-specific type. Stock llama.cpp will refuse to load it. Requires a ROCmFPX fork build — see Runtime. Upstream charlie12345/ROCmFPX loads these files too (any build with the custom GGML types).
Best served on the Vulkan backend (RADV). The Q6 ROCmFPX staging path is optimized in the Vulkan build of the fork: on the ROCm backend (older commits) pp512 drops ~50%. On Vulkan RADV: 1120 pp512 / 49.2 tg128.
AMD RDNA 3.5 (gfx1151 / Strix Halo) is the target. Tested on Radeon 8060S iGPU. Not tested elsewhere.
On Strix Halo, GGML_CUDA_ENABLE_UNIFIED_MEMORY=1 (ROCm) or the Vulkan backend with default ICD is required — VRAM partition is 512 MB.

Benchmarks

Hardware: AMD Ryzen AI Max+ 395, Radeon 8060S (gfx1151, 128 GB unified LPDDR5X). Method: llama-bench -ngl 999 -fa on -p 512 -n 128, Vulkan RADV backend (fork build).

Format	Size	tg128 (tok/s)	pp512 (tok/s)
Q6_0_ROCMFPX (this)	27.39 GiB	49.2	1120
Q4_0_ROCMFP4_STRIX_LEAN (same base)	17.73 GiB	81.6	1164.7
UD-Q5_K_M_MTP (unsloth)	25.2 GiB	57.9	1008.1

Q6_0_ROCMFPX trades decode speed for quality: vs the fp4 preset it is ~-39% tg, but at ~6.5 BPW it sits in Q6_K-class quality territory (community K-quant comparisons put Q6_0_ROCMFPX at Q6_K lossless level, with +30% pp over Q6_K on the fork's optimized paths).

MTP (speculative decoding)

With --spec-type draft-mtp --spec-draft-ngl all --spec-draft-p-min 0.0 --spec-draft-p-split 0.10 (config validated by the fork's regression guard for qwen35-a3b). n-max sweep on this file (server-timing, 2 prompts × 2 runs, ctx 16k, Vulkan RADV):

n-max	prose tg (tok/s)	deterministic tg (tok/s)
4	44.7	62.8
3 (recommended)	54.0	67.3
2	58.2	62.4

Recommendation: n-max 3 for mixed/agentic workloads (best deterministic content, balanced prose); n-max 2 for pure decode. Position-4 acceptance is only ~0.51 on this base model, so the extra draft cost of n-max 4 is not paid back. Acceptance pos-1 0.82–0.87, mean accepted length 2.5–3.7 tokens.

Comparison — every Qwen3.6-35B-A3B variant we tested

All rows: same host (Ryzen AI Max+ 395, 8060S, 128 GB), same method (llama-bench -ngl 999 -fa 1 -p 512 -n 128), 2026-08-14.

Variant	Quant	Size	Backend	pp512	tg128
base — this repo	Q6_0_ROCMFPX	27.39 GiB	Vulkan RADV	1120.4	49.2
Qwopus3.6-35B (abliterated fine-tune)	Q6_0_ROCMFPX	27.38 GiB	Vulkan RADV	1076.9	49.5
Qwopus3.6-35B (abliterated fine-tune)	Q6_0_ROCMFPX	27.38 GiB	ROCm	520.6 ⚠️	51.3
base	Q4_0_ROCMFP4_STRIX_LEAN	17.73 GiB	Vulkan RADV (fork)	1164.7	81.6
base	Q4_0_ROCMFP4_STRIX_LEAN	17.73 GiB	ROCm	1420.7	71.2
base	UD-Q5_K_M_MTP (unsloth)	25.22 GiB	Vulkan RADV	1008.1	57.9
base	UD-Q5_K_M_MTP (unsloth)	25.22 GiB	ROCm	1359.4	50.8
base	UD-Q5_K_M_MTP (unsloth)	25.22 GiB	Vulkan AMDVLK	663.0	55.8

Notes:

"Vulkan RADV (fork)" = the charlie12345/ROCmFPX fork built Vulkan-only — the only Vulkan build that runs the ROCmFPX tensor types (stock llama.cpp rejects them).
⚠️ pp 520 on ROCm: the Q6 staging path is unoptimized in the ROCm build — for Q6_0_ROCMFPX the reference backend is Vulkan.
tg128 leaderboard: fp4 LEAN on the Vulkan fork (81.6) > fp4 on ROCm (71.2) > UD-Q5_K_M on RADV (57.9) > Q6 (49.2). pp leaderboard: ROCm fp4 (1420.7) > RADV fp4 (1164.7) > Q6 RADV (1120.4). The Q6 is the quality tier, not the speed tier.
Our other Strix Halo ROCmFPX repos (grug-35b-v2, Ornith-1.0-35B, Nemotron-Lightning-30B) were benchmarked in ROCm containers (HIP backend) — do not compare their tg/pp numbers directly against the Vulkan rows above.

Usage (Vulkan, fork build)

bash

llama-server -m Qwen3.6-35B-A3B-MTP-Q6_0_ROCMFPX.gguf \
  -ngl 999 -fa on --jinja --parallel 4 \
  --top-p 0.95 --top-k 20 --temperature 1.0 \
  --reasoning on --reasoning-budget 16384
# + MTP: --spec-type draft-mtp --spec-draft-ngl all --spec-draft-n-max 3

Note: without --reasoning-budget, thinking defaults to unlimited (INT32_MAX) and can run away.

Quantization details

Type: Q6_0_ROCMFPX — fork-specific (~6.5 BPW).
Source: BF16 GGUF from unsloth/Qwen3.6-35B-A3B-MTP-GGUF, quantized with llama-quantize from the fork (imatrix pass attempted; final config in repo history).
MTP layers retained (~0.8 GiB extra): the file can be served with or without speculative decoding; MTP is off unless --spec-type draft-mtp is passed.
Base model: Qwen/Qwen3.6-35B-A3B (Apache-2.0).

License

Apache-2.0 (inherited from the base model). See LICENSE.

Runtime

Benchmark environment: one bare-metal AMD Strix Halo (Ryzen AI MAX+ 395, 128 GB) — full dated configuration and measurement policy: BARE-METAL.md

Requires a ROCmFPX fork build (custom tensor types — stock llama.cpp refuses the file). Recommended: our lab build (pugant/strix-nebulosa, main) — reasoning budget and persistent prompt cache on every model; drafter features where the model ships one: see the engine section of its README.

Runs on any build with the Q6_0_ROCMFPX type (MTP speculative decoding included in the GGUF); our lab build adds reasoning budget + persistent prompt cache.

Everything here is experimental and provided as-is, at your own risk.

pugant/Qwen3.6-35B-A3B-MTP-Q6_0_ROCMFPX

Qwen3.6-35B-A3B — Q6_0_ROCMFPX (Strix Halo / gfx1151, Vulkan)

TL;DR