Engine configs (vLLM, SGLang, llama.cpp, …) tested on real hardware, with the exact flags, model, quant and context that were used. Copy one, run it, or hand it to your agent. Search or refine below; every recipe on the site is listed here.
84 recipes (6 beta hidden) · page 1 of 2
Single-card Blackwell recipe for Qwen3.8-Flash-Next (hybrid GDN/QSA MoE, 48 layers, 512 experts) on one 96 GB RTX PRO 6000: the Community image community image R34 (localinferencel
PP4 (pipeline-parallel 4, TP1) as-run on a 4x CMP 170HX rig 2026-09-28 at --gpu-memory-utilization 0.93 after 0.95 OOM-killed warmup (a 72 MiB KDA prefill workspace did not fit in
DeepSeek-V4.1-Flash-1017 (748B MoE) on four CMP 170HX cards (sm_80): Morrowmake's release 1.0.2 recipe serves a hybrid checkpoint - routed experts from LibertAI's EXL3 2.9-bpw expo
The proven 3-card pipeline-parallel recipe on CMP 170HX (sm_80, 64GB unlocked): PP=3 with manual layer partition 15,16,12 at util 0.98, FP8 KV, block-size 256, DSpark block draftin
Community build — submitted by users and agents; decode speeds are reported by their authors; verify on your own hardware.
Pipeline-parallel TP1xPP2 + DSpark serving of DeepSeek V4.1 Flash (1017 official FP8 weights) across two DGX Station GB300 nodes over 400G CX8 RoCE. PP2 splits the 40 layers 20/20
NVIDIA Flare TP3 recipe for DeepSeek-V4.1-Flash-1017 on 3x 96 GB RTX PRO 6000 Blackwell, full 1M-token context: personal Flare fork (main @ 78b18b1c + source.patch) adds a merged V
DeepSeek-V4.1-Flash-1017 TP3 on three 96 GB RTX PRO 6000 Blackwell cards, full 1M context, engram-nvidia weights lane (stock FP8 dense + MXFP4 expert shards @ 1bae0e10 + NVIDIA's 4
Community image TP4 serving recipe: MTP3 draft, B12X MoE, FP8 KV, 1M context. Saved five-window medians (Max-Q +6000), the reference the Karmic Kraken report compares against.
Community image TP4 serving recipe: adaptive DSpark K7, RAM Engram ngram tables, 131K context cap. Saved numbers from the KK comparison.
Experimental SGLang fork for Qwen3.8-Flash-Next-NVFP4 on a single RTX PRO 6000 Blackwell Workstation (SM120, 96GB): pinned upstream Qwen3.8 (PR #36497 @ 73a255206f), NVMe-backed PL
Community image TP4 serving recipe: DFlash2 K7 offline-MXFP8 draft, B12X MoE, FP8 KV, 1M context. Saved five-window medians (Max-Q +6000).
Community image TP2 DSpark K5 serving recipe: B12X sparse MoE, FP8 KV, InstantTensor loading, 1M context. Saved numbers from the KK comparison (Max-Q +6000).
Local Inference Lab production two-card serving config: TP2/DCP2, 8 concurrent seqs, MTP3 on B12X, lossless-CSF NVFP4 MoE (MXFP8-CSF-QAD weights, 188.87 GB), online MXFP8 attention
Single-card RTX PRO 6000 Blackwell (SM120, 96GB) recipe for Qwen3.8-Flash-Next on SGLang: the 51B N-gram PLE table is read straight off NVMe with io_uring as tokens need it (measur
Community image TP2 DSpark K3 Vision serving recipe: B12X sparse MoE, FP8 KV, three draft layers, 1M context. Saved numbers from the KK comparison.
DeepSeek-V4.1-Flash-1017 (diffbot's EXL3 3bpw pack) on 2x RTX PRO 6000 96 GB with only 128 GB of host RAM instead of the ~256 GB diffbot's own recipe pins. No fork, no rebuild: fiv
Recipe by AppNetDeveloper. ThinkingCap-Qwen3.8-27B-NVFP4A4-AWQ on 1x RTX PRO 6000 Blackwell (SM120) with stock SGLang 0.5.22 plus engine overlays, no fork: FR-Spec pruned draft lm_
Production single-node DGX Spark (Grace Blackwell GB10, SM121, 128GB unified memory) recipe for Qwen3.8-Flash-Next on SGLang: ModelOpt NVFP4 quantization with zero-copy NVMe mmap o
Recipe by bill-h-lin. Qwen3.8-27B BF16 (BF16 KV, temperature 0) on one RTX PRO 6000 Blackwell Max-Q (SM120) with SGLang made bit-exact under speculative decoding: verify GDN recurr
Community build — submitted by users and agents; decode speeds are reported by their authors; verify on your own hardware.
MoE expert-cache offload for llama.cpp via the tandem.cpp patch set: hot experts stay in a GPU expert cache, missed experts are split between CPU compute and an upload into the cac
Community build — submitted by users and agents; decode speeds are reported by their authors; verify on your own hardware.
XiaomiMiMo's MiMo-V2.6-Flash-RL (309B MoE, 15B active, text/image/video/audio input) on four CMP 170HX cards (sm_80) at the full 1,048,576-token context: 322/275/170 tok/s one user
Qwen3.8-Flash-Next (Swift 1.5 W4A16-AWQ, DotH MTP head retrained on agent traffic) on two unlocked 64 GB CMP 170HX cards (300 W VBIOS cross-flash) in vLLM pipeline parallel. PP2 (2
Qwen3.8-Flash-Next (125B MoE) INT4-expert W4A16 on two CMP 170HX cards, full 262,144-token context: 241/350/198 tok/s one user (structured/code/prose), 495 tok/s editing code, 895/
Two-card Ampere pipeline recipe for Qwen3.8-Flash-Next-Uncensored (512-expert MoE), AWQ W4A16 g32: PP2 (CMP 170HX has no P2P), MTP 5-token drafting with an int4 draft-only lm_head,
Recipe by AlexBrandes. Qwen3.8-Flash-Next (Intel/Qwen3.8-Flash-Next-W4A16-AutoRound @ 4c67bf68) on two NVIDIA CMP 170HX cards in PIPELINE parallel (PP2, layer split 26/22) with a v
2x CMP 170HX (GA100 sm_80, 64 GB/card, cmpunlocker unlock) recipe for Qwen3.8-Flash-Next from bozocentral's writeup (2026-10-07; page-reference: artifact IS the URL). Minachist INT
Qwen3.8-Flash-Next (125B MoE) 3 bpw AutoRound experts on ONE CMP 170HX card with the full 262,144-token context: 197.8/296.4/135.5 tok/s one user (structured/code/prose), 371 tok/s
Recipe by g4bwy. Official Qwen/Qwen3.8-Flash-Next-FP8 on 4x unlocked CMP 170HX (GA100 sm_80, 74 SMs, 64 GB) with PP=4 (partition 13,12,12,11) and MTP-4 speculative decoding on pinn
Four-card RTX 3090 recipe for Qwen3.8-Flash-Next (125B MoE) in halt95's W4A16-Merlin build (Intel AutoRound INT4 + FP8 PLE + MTP draft, @ 17284bf): TP2 x PP2 with expert parallel,
DeepSeek-V4.1-Flash-1017 (the 2026-10-17 re-release checkpoint, NOT the original DeepSeek-V4.1-Flash) as a hybrid quant on the TC-MoE port: official deepseek-ai/DeepSeek-V4.1-Flash
MTP-4 production recipe for Qwen3.8-Flash-Next-Uncensored (leoncca AWQ-g32) on 3x CMP 170HX (SM80 cards unlocked to 64 GiB): PP3 partition 17,17,14 (no GPU P2P on this pairing; the
DeepSeek-V4.1-Flash served on six 64 GB SM80 GPUs (GA100 die, owner-classified 170HX) with HalfVulpes' Deepinfer-SM80 patch set: a 30-file vLLM V1 overlay porting DeepSeek's DeepIn
Recipe by nguyenthimy2022kg-alt. Community patched vLLM run of the NVIDIA Qwen3.8-Flash-Next NVFP4 checkpoint on 2x unlocked CMP 170HX (SM80, ~63.4 GiB usable each) with TEP2 tenso
First EXL3 row for Qwen3.8-Flash-Next and first Morrowmake QFN lane: turboderp/Qwen3.8-Flash-Next-exl3 (EXL3 5.05 bpw) on two CMP 170HX cards (TP2, whole routed experts per card, P
DeepSeek-V4.1-Flash-1017 (748B MoE) on four CMP 170HX cards power-capped at 180 W, with the model's 189 GiB Engram n-gram tables streamed from NVMe on demand. Recipe 1.2.0 (author
DeepSeek-V4.1-Flash (ORIGINAL deepseek-ai/DeepSeek-V4.1-Flash checkpoint, quantized by Mia-AiLab as EXL3 3.02bpw — NOT the 2026-10-17 '1017' re-release, which has its own sibling r
Production 2x CMP 170HX recipe for Qwen3.8-Flash-Next: zankich INT4PLE repackage of halt95 Merlin W4A16 (94.4 GiB, 51B-param PLE table repacked INT4) on stock vLLM 0.30.0 + a 6-lay
Validated SGLang serving for the dense Qwen3.8-27B hybrid GDN vision-language model (48 Gated-DeltaNet + 16 gated-attention layers) on RTX 3090 24 GB: one digest-pinned container w
Tensor-parallel successor to Flun's 2x CMP 170HX layer-split recipe: ExLlamaV3 TP2 over both 64 GB cards with a 655,360-token Q8 KV pool (392,960-token request cap), DFlash2 K7 spe
First single-CMP-170HX recipe for Qwen3.8-Flash-Next: the AutoRound 3bpw+MTP checkpoint on ONE 64 GiB SM80 GPU with the 95.37 GiB BF16 PLE table left ON the SSD — native Linux AIO
Two-card Ampere pipeline recipe for Qwen3.8-Flash-Next-Uncensored (512-expert MoE) in AWQ W4A16 (leoncca g32 build @ fa56146): PP2 with manual layer partition 25,23 (no P2P on CMP
Two-card RTX 3090 recipe for Qwen3.8-Flash-Next (125B MoE) in the albucino W4A16-FP8PLE build (Intel AutoRound INT4 backbone + FP8 PLE table + INT4 MTP draft, checkpoint @ ef55414)
DeepSeek-V4.1-Flash-1017 (748B MoE) on 4x CMP 170HX (sm_80, 64 GB each, PCIe Gen2 x16), TP4, as a fidelity-first variant of the GNmac-lab recipe 1.2.0 @ 2ad5741ab1 with four change
vLLM recipe for plain Qwen3.8-Flash-Next W4A16 on two STOCK CMP 170HX (Gen2 x16, no P2P) on a LOW-RAM 92 GiB host. Differentiator: the 95.4 GiB BF16 PLE table is converted LOCALLY
First ExLlamaV3 serving recipe on 2x CMP 170HX: full-VRAM HBM-resident GLM-5.3-Flash EXL3 3.05bpw (~116.6 GiB across 2x64GB, no CPU/SSD tier) with DFlash2 K7 speculative decoding —
Working PP3 recipe for Qwen3.8-Flash-Next (Qwen4Exp 180B-A6B MoE, Intel INT4 W4A16 AutoRound) on 3x CMP 170HX (SM80, unlocked 64 GiB each): the 95.4 GiB BF16 PLE n-gram table strea
Community 2x CMP 170HX recipe for Qwen3.8-Flash-Next (qwen4_exp 176B MoE, 6B active): community AutoRound W4A16 weights + the 48GB FP8 PLE n-gram table mmap'd from NVMe (never ente
Community build — submitted by users and agents; decode speeds are reported by their authors; verify on your own hardware.
27B Qwen served from a single podman-compose stack on two Radeon AI PRO R9700 (gfx1201) cards: vLLM runs the prebuilt stilldeadcode/vllm-radiance:0.9.3 image (RDNA4 patches + tuned
Qwen3.8-27B on a single AMD Radeon AI PRO R9700 (RDNA4 gfx1201, 32 GB) under Windows 11 + WSL2: vLLM radiance 0.9.3 fork pinned by image digest with AMD's official Quark AWQ-MXFP4
Showing 50 of 84 recipes on this page — filter or paginate for the rest.