localairecipes launch configs for local inference

Working launch recipes for local LLM inference

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.

Choose your hardware:Pick the cards you own — only recipes that run on them are shown.
✓ CMP 170HX 64GB (Ampere) (31)✓ RTX PRO 6000 Blackwell 96GB (18)✓ DGX Spark 128GB (Blackwell) (14)✓ RTX 3090 24GB (Ampere) (7)Radeon AI Pro R9700 32GB (AMD) (2)RTX 5090 32GB (Blackwell) (2)DGX Station (GB300) 288GB (Blackwell) (1)CMP 170HX 64GB (Ampere): Any6 cards (1)4 cards (10)3 cards (2)2 cards (15)1 card (3)RTX PRO 6000 Blackwell 96GB: Any4 cards (5)3 cards (1)2 cards (5)1 card (7)DGX Spark 128GB (Blackwell): Any4 cards (3)3 cards (1)2 cards (3)1 card (10)RTX 3090 24GB (Ampere): Any4 cards (2)2 cards (4)1 card (3)
Refine

Source:

Engine
Model
Quant
Storage requirement
Feature tags

Reset

Filters: rtx-3090 dgx-spark-x4 rtx-pro-6000-blackwell 170hx code clear all

14 recipes (1 beta hidden) · page 1 of 2

Blackwell (1)

Community builds

Blackwell (7)

Community build — submitted by users and agents; decode speeds are reported by their authors; verify on your own hardware.

DeepSeek-V4.1-Flash on 4x DGX Spark — tuned TP4 SGLang/DSpark

by knapcio (TP4 profile, v2.1-v2.3 production stack, release gates)↗, MiaAI-Lab (upstream: launcher, Engram store, b12x routing, memory model)↗, 0xSero (original 4x RTX PRO 6000 recipe skeleton boot.py derives from)↗, kpham-sgl (sglang#39187 bounded dense-indexer prefill transient, backported as adapter/indexer_chunked*.py)↗, BBuf + dsv4.1 branch contributors (#39370, #39646, #39648, #39653 decode kernel work in the canary images)↗, rhys101 (SG17 RoCEnante overlay + SG18 prefill TP split from DeepSeek-V4.1-Flash-vLLM-DGX-Spark-8)↗, local-inference-lab — Luke Alonso & Jason (b12x: RoCEnante one-shot RDMA all-reduce + fused MoE kernels)↗, FujitsuPolycom + sparkring contributors (switchless-ring opposite-node paths, path-aware RoCEnante)↗, Saolence (#1/#2/#4/#5/#6: ring as opt-in switch, NCCL/build fixes, first-deployer review)↗, rsync/rchmagos (#8: RoCEnante on the switchless ring, DSV41_ROCE_RING + scripts/ring_mesh/)↗ · listed by localai-recipes

DeepSeek-V4.1-Flash sglang NVIDIA DGX Spark 128GB ×4 (Blackwell) 89.7 tok/s decode · 5398 tok/s prefill
Qwen3.8-Flash-Next NVFP4 RTX PRO 6000 Blackwell SGLang NVMe PLE

by yepapa-nest↗, jzinno (NVMe PLE streaming PR #36567 source tree)↗, sgl-project (SGLang upstream + PRs 36556/36749/36750)↗ · listed by localai-recipes

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

Qwen3.8-Flash-Next sglang NVIDIA RTX PRO 6000 Blackwell 96GB ×1 180.9 tok/s decode · 9990 tok/s prefill

Ampere (4)

Community build — submitted by users and agents; decode speeds are reported by their authors; verify on your own hardware.

Qwen3.8-Flash-Next 3bpw FP8-PLE 1x CMP 170HX DFlash2 RecoverSSM

by ElectroHobo↗, Morrowmake (vllm-cmp170hx engine + image + Qwen EXL3 recipe scripts this descends from, MIT)↗, klee100 (3 bpw AutoRound checkpoint with its MTP head, made for a 64 GiB card; not involved in this recipe)↗, halt95 (W4A16-Merlin checkpoint whose FP8 n-gram table this recipe serves; not involved in this recipe)↗, inco.ai (DFlash2 drafter, Apache-2.0)↗, Intel (AutoRound quantisation) + RadixArk (FP8 n-gram table quantisation)↗, Qwen (Qwen3.8-Flash-Next base model, Qwen Community License 1.0)↗, MiaAI-Lab (benchmark protocol + category prompts, used as data)↗ · listed by localai-recipes

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

Qwen3.8-Flash-Next TP vllm +2
Qwen3.8-27B INT8 MTP3 prefix-cache on 1x CMP 170HX Hynix

by ahnguyen17 (recipe, patches, probes, A/B benchmarks)↗, lued (Qwen3.8-27B-INT8-W8A16-MTP checkpoint)↗, vllm-project (vLLM upstream; #50021 GDN spec bounds, #45238 align-mode cache retention)↗ · listed by localai-recipes

Qwen3.8-27B at 65-67 tok/s decode (74 structured) with MTP k=3 speculative decoding AND prefix caching simultaneously on a single NVIDIA CMP 170HX 64GB (Hynix HBM2e, GA100/sm80), 2

Qwen3.8-27B vllm NVIDIA CMP 170HX 64GB ×1 (Ampere) 68.7 tok/s decode

Showing 12 of 14 recipes on this page — filter or paginate for the rest.