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

GLM-5.3-Flash (320B MoE, 18B active) (1)

EXL3 3.05 bpw: turboderp/GLM-5.3-Flash-exl3; DFlash2 drafter converted to EXL3 6 bpw (1)

ExLlamaV3 (patched, pinned) (1)

upstream dev 0f685aa9fd plus 9 patches; repo db01c256 (1)

8 GB · Other · None

2 (1)

TP2 over NCCL (GPU P2P), DFlash2 K7 plus hybrid drafting (1)

392,960 per request; 655,360-token Q8 KV pool shared by two requests (1)

107.67 (1)

1,533 (1)

about 62.9 / 62.3 GiB idle, peaks about 64.4 / 63.0 GiB with vision (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

610.57.04 (1)

cmpunlocker (64 GiB); P2P-capable driver for the NCCL backend (1)

Reported (1)

nestaa51 (1)

October 13, 2026 (1)

https://github.com/alexvnesta/glm53-flash-cmp170hx-tp (1)