Drilldown: Recipe

DeepSeek V4.1 Flash heretic on 4x 8 GB · DeepSeek V4.1 Flash on 2x 8 GB (Project Maya) · DeepSeek V4.1 Flash on 4x 8 GB (alanoo fidelity build) · DeepSeek V4.1 Flash on 4x 8 GB (CurtisAccelerate TC-MoE) · DeepSeek V4.1 Flash on 4x 8 GB (GNmac) · DeepSeek V4.1 Flash on 4x 8 GB (Morrowmake) · DeepSeek V4.1 Flash on 4x 8 GB plus RTX PRO 6000 (ExLlamaV3 TP5) · DeepSeek V4.1 Flash on 6x 8 GB (Deepinfer-SM80 TP2 x PP3) · DeepSeek V4.1 Flash on 6x 8 GB (TP6) · DeepSeek V4.1 Flash on 8x 8 GB (Deepinfer-SM80 TP4 x PP2) · DeepSeek V4.1 Flash on 8x 8 GB (Zanooda PP8) · GLM 5.3 Flash on 2x 8 GB (ExLlamaV3 layer split) · GLM 5.3 Flash on 2x 8 GB (ExLlamaV3 TP2) · GLM 5.3 Flash on 2x 8 GB (Project Maya) · GLM 5.3 Flash on 3x 8 GB · GLM 5.3 Flash on 4x 8 GB · GLM 5.3 on 8x 8 GB (Int4-Int8Mix PP8) · GLM 5.3 on 8x 8 GB (vllm-170hx-glm5) · MiMo V2.6 Flash on 4x 8 GB · Qwen3.8 27B on 1x 8 GB (W4A16 vLLM) · Qwen3.8 Flash Next on 1x 8 GB (ElectroHobo 3 bpw) · Qwen3.8 Flash Next on 1x 8 GB (iIIusi0n SSD PLE) · Qwen3.8 Flash Next on 2x 8 GB (DUFRelic PP2) · Qwen3.8 Flash Next on 2x 8 GB (ElectroHobo W4A16 TP2) · Qwen3.8 Flash Next on 2x 8 GB (mannydelrio PP2) · Qwen3.8 Flash Next on 2x 8 GB (Morrowmake EXL3) · Qwen3.8 Flash Next on 2x 8 GB (Plean PP2) · Qwen3.8 Flash Next on 2x 8 GB (vektorprime PP2) · Qwen3.8 Flash Next on 3x 8 GB (clee PP3) · Qwen3.8 Flash Next on 4x 8 GB · Other · None

Qwen3.8-Flash-Next (125B MoE per the recipe) (1)

W4A16: halt95/Qwen3.8-Flash-Next-W4A16-Merlin (AutoRound INT4 g128 experts, INT8 GDN projections, FP8 n-gram table) (1)

Morrowmake/vllm-cmp170hx image (Qwen EXL3 release) with patch files mounted read-only (1)

recipe 0.3.1; engine 18305948ac (1)

8 GB · Other · None

2 (1)

TP2 with whole-expert expert parallel (1)

262,144 (1)

349.7 (1)

4,696 (1)

64 GiB, GPU_UTIL 0.95 (fixed) (1)

None · 180 W · 200 W · 140 W · 140 W (200 W observed-draw abort guard) · 150 W per CMP (PRO 400 W) · 170 W (author's 10-card run) · 180 W (author's cards); about 170 W with a modest memory OC in his Discord numbers · 180 W (nvidia-smi -pl 180) · 180 W (one card unstable at 250 W) · Other

580 or newer (measured on 610.57.04); open modules with HMM for the table-on-disk mode (1)

cmpunlocker (64 GiB, 74 SMs) with a P2P-capable build (1)

Proven (1)

nathang6866 (1)

October 23, 2026 (1)

https://github.com/ElectroHobo/qwen38-flash-next-w4a16-2x-cmp170hx-recipe (1)