Drilldown: Recipe

DeepSeek V4.1 Flash on 4x 8 GB (GNmac) (1) · GLM 5.3 Flash on 4x 8 GB (1) · Qwen3.8 Flash Next on 1x 8 GB (iIIusi0n SSD PLE) (1) · Qwen3.8 Flash Next on 2x 8 GB (ElectroHobo W4A16 TP2) (1)

DeepSeek-V4.1-Flash-1017 (1) · GLM-5.3-Flash (320B MoE, 18B active) (1) · Qwen3.8-Flash-Next (1) · Qwen3.8-Flash-Next (125B MoE per the recipe) (1)

AutoRound 3 bpw + MTP: klee100/Qwen3.8-Flash-Next-AutoRound-3bpw-MTP; 95.37 GiB BF16 PLE table on SSD (1) · Hybrid: routed experts EXL3 (mostly 3 bpw, 2 bpw on layers 18-22) from LibertAI's 2.9 bpw pack; attention, shared experts, indexer and DSpark drafter official FP8; KV fp8_ds_mla (1) · W4A16 (INT4 weights, group size 128, FP16 activations): canada-quant/GLM-5.3-Flash-W4A16-MTP (1) · W4A16: halt95/Qwen3.8-Flash-Next-W4A16-Merlin (AutoRound INT4 g128 experts, INT8 GDN projections, FP8 n-gram table) (1)

Morrowmake/vllm-cmp170hx (vLLM fork, container image) (1) · Morrowmake/vllm-cmp170hx image (Qwen EXL3 release) with patch files mounted read-only (1) · vLLM a5a30471 plus a 13-file downstream patch (native Linux AIO PLE reader) (1) · vLLM with a pinned sm_80 patch stack (built locally from source) (1)

0.29.1rc1.dev402+ga5a30471f.ple1 (1) · recipe 0.3.1; engine 18305948ac (1) · recipe 1.2.0 (1) · recipe 1.8.0, engine 819146023f (1)

8 GB (4)

1 (1) · 2 (1) · 4 (2)

single card (1) · TP2 with whole-expert expert parallel (1) · TP4 (default) or PP4 (1) · TP4 (default) or PP4 for x1/x4 links (1)

262,144 (2) · 262,144 (up to 1,048,576) (1) · 262,144 advertised (tested through 32K) (1)

110.99 (1) · 258 (1) · 349.7 (1) · 442.1 (1)

2,416 (1) · 2,636 (1) · 3,161 (1) · 4,696 (1)

64 GiB exposed, --gpu-memory-utilization 0.95 (1) · 64 GiB, --gpu-memory-utilization 0.96 (1) · 64 GiB, GPU_MEMORY_UTILIZATION 0.95 (1) · 64 GiB, GPU_UTIL 0.95 (fixed) (1)

180 W (3) · 180 W (nvidia-smi -pl 180) (1)

None · 580 or newer (measured on 610.57.04) · NVIDIA 580 or newer; P2P needs the Morrowmake/cmpunlocker build · 580 or newer (measured on 610.57.04); open modules with HMM for the table-on-disk mode · 610.43.02 (nvidia-driver-610-open) · 610.43.03 · 610.43.03 (open modules) · 610.57.04 · CUDA 13 capable; measured on 610.43.02 · NVIDIA 610.43.02 · NVIDIA open 610.43.02 from a cmpunlocker fork with a P2P patch · NVIDIA open 610.43.03, CUDA 13.0 · NVIDIA open 610.57.04 patched with Morrowmake/cmpunlocker 1f2c498 (profile 8gb) · nvidia-open 610.43.02 + cmpunlocker v0.5 or newer · Other

cmpunlocker (64 GiB) (1) · cmpunlocker (64 GiB, 74 SMs) with a P2P-capable build (1) · cmpunlocker (64 GiB, 74 SMs) with a P2P-capable driver for TP4 (1) · cmpunlocker (64 GiB, 74 SMs); Morrowmake/cmpunlocker for P2P (1)

Reported · Proven · Other · None

greencandle9753 (1) · iIIusi0n (1) · nathang6866 (1) · voidgiraffe (1)

September 19, 2026 (1) · October 16, 2026 (1) · October 23, 2026 (1) · October 26, 2026 (1)

https://github.com/ElectroHobo/qwen38-flash-next-w4a16-2x-cmp170hx-recipe (1) · https://github.com/GNmac-lab/deepseek-v41-flash-1017-cmp170hx (1) · https://github.com/iIIusi0n/qwen3.8-flash-next-cmp170hx (1) · https://github.com/Morrowmake/glm53-flash-cmp170hx-recipe (1)