Local AI Registry

Recipe

laguna-s-2-1-nvfp4-rtx-5090-32gb-vllm-tp4

laguna-s-2-1-nvfp4-rtx-5090-32gb-vllm-tp4
JSON APIFind compatible recipes

Observed LocalMaxxing leaderboard run. Evidence for compatibility, not an executable launch contract.

Record

Status
candidate
Source
localmaxxing
Engine
vllm
Engine version
0.23.1rc1.dev-nightly
Accelerators
4
Tensor parallel
4
Context tokens
262,144
Max concurrency
1
chat
unknown
reasoning
unknown
tools
unknown
vision
unknown

Hugging Face model card

Identity

https://huggingface.co/poolside/Laguna-S-2.1-NVFP4
Repository
poolside/Laguna-S-2.1-NVFP4
Status
known
Link type
Exact Hub repository

Public Hugging Face repository confirmed by the Hub API.

Candidate: useful compatibility or speed evidence. The registry does not offer Run until promotion requirements are met.

Observed configuration

vllm

Evidence only · candidate · reference

Candidate evidence — not a Run contract

Source
https://www.localmaxxing.com/en/runs/cmrxk8q0x00dco401yhwzjx9m

Observed source tokens

Mechanical split of the source command. Unverified against the engine CLI. The registry does not offer Run for this recipe.

  1. docker
  2. run
  3. --gpus
  4. all
  5. --ipc=host
  6. --network=host
  7. --shm-size=32g
  8. -e
  9. CUTE_DSL_ARCH=sm_120a
  10. -v
  11. <MODEL_DIR>:/model:ro
  12. vllm/vllm-openai:<nightly-cu13>
  13. --model
  14. /model
  15. --served-model-name
  16. laguna
  17. --trust-remote-code
  18. --tensor-parallel-size
  19. 4
  20. --disable-custom-all-reduce
  21. --dtype
  22. bfloat16
  23. --kv-cache-dtype
  24. fp8
  25. --max-model-len
  26. 262144
  27. --gpu-memory-utilization
  28. 0.90
  29. --max-num-seqs
  30. 1
  31. --max-num-batched-tokens
  32. 8192
  33. --enable-prefix-caching
  34. --override-generation-config
  35. {"temperature":0.7,"top_p":0.95}
FlagValue
--gpusall
--ipchost
--networkhost
--shm-size32g
-eCUTE_DSL_ARCH=sm_120a
-v<MODEL_DIR>:/model:ro
--model/model
--served-model-namelaguna
--tensor-parallel-size4
--dtypebfloat16
--kv-cache-dtypefp8
--max-model-len262144
--gpu-memory-utilization0.90
--max-num-seqs1
--max-num-batched-tokens8192
--override-generation-config{"temperature":0.7,"top_p":0.95}

Measured speed

ConcurrencyContextPrefillDecodeTTFT msStatusSweep
1262,1446,262157.1—observedlaguna-s-2-1-nvfp4-rtx-5090-32gb-vllm-tp4-sweep

Remaining fields

Identity, launch, related records, and measured speed are shown above. This is the rest of the normalized record.

hardware count
4
hardware id
rtx-5090-32gb
id
laguna-s-2-1-nvfp4-rtx-5090-32gb-vllm-tp4
model instance id
poolside-laguna-s-2-1-nvfp4--nvfp4
recipe source
localmaxxing
schema version
local-ai-registry/v1
speed sweep ids
laguna-s-2-1-nvfp4-rtx-5090-32gb-vllm-tp4-sweep
status
candidate

capabilities

draft launch

accelerator backend
nvidia
arguments
--model, poolside/Laguna-S-2.1-NVFP4, --tensor-parallel-size, 4, --host, 0.0.0.0, --port, 8000, --max-model-len, 262144
container port
8,000
host port
8,000
image
vllm/vllm-openai@sha256:0a51ea5b4ae2dc5d81890e5173f54203d2a3ae0cfffe51b8fd2afd4391bfd967
ipc
host
kind
docker
shm size
16g
synthesized · generated at
2026-08-31T22:12:17Z
synthesized · image provenance
deepseek-fp8-rtx-pro-6000-blackwell-96gb-vllm-tp1
synthesized · template
vllm-openai-v1

mounts

read onlytarget
No/root/.cache/huggingface

engine

name
vllm
version
0.23.1rc1.dev-nightly

serving

max concurrency
1
max context tokens
262,144
tensor parallel
4
Provenance & metadata (3)

facts

capabilities.chat · provenance · captured at
2026-08-30T09:10:02Z
capabilities.chat · reason capability-not-verified
capabilities.chat · state unknown
capabilities.reasoning · provenance · captured at
2026-08-30T09:10:02Z
capabilities.reasoning · reason capability-not-verified
capabilities.reasoning · state unknown
capabilities.tools · provenance · captured at
2026-08-30T09:10:02Z
capabilities.tools · reason capability-not-verified
capabilities.tools · state unknown
capabilities.vision · provenance · captured at
2026-08-30T09:10:02Z
capabilities.vision · reason capability-not-verified
capabilities.vision · state unknown
engine.graph mode · provenance · captured at
2026-08-30T09:10:02Z
engine.graph mode · reason runtime-detail-not-published
engine.graph mode · state unknown
serving.kv cache tokens · provenance · captured at
2026-08-30T09:10:02Z
serving.kv cache tokens · reason kv-cache-capacity-not-published
serving.kv cache tokens · state unknown

metadata

localmaxxing · hardware label
RTX 5090
localmaxxing · notes
poolside Laguna-S-2.1-NVFP4 (117.6B MoE / 8.5B active, 256 experts, 36 SWA@512 + 12 global layers) on 4x RTX 5090 32GB, tensor-parallel 4 (no-NVLink, --disable-custom-all-reduce), no speculative decode. FP8 KV cache, FlashInfer, CUDA graphs, prefix caching. b1 = pure single-stream decode over 1024 forced tokens (ignore_eos), 3 runs temp 0.7 / top_p 0.95: 157.02 / 157.13 / 157.18 tok/s (mean 157.11, stdev 0.07); TTFT ~5.1s over the 32k prompt (tokSPrefill from prompt/TTFT, tokSTotal = (prompt+output)/end-to-end). Full 262144 context fits at TP=4.
localmaxxing · observed command
docker run --gpus all --ipc=host --network=host --shm-size=32g -e CUTE_DSL_ARCH=sm_120a -v <MODEL_DIR>:/model:ro vllm/vllm-openai:<nightly-cu13> --model /model --served-model-name laguna --trust-remote-code --tensor-parallel-size 4 --disable-custom-all-reduce --dtype bfloat16 --kv-cache-dtype fp8 --max-model-len 262144 --gpu-memory-utilization 0.90 --max-num-seqs 1 --max-num-batched-tokens 8192 --enable-prefix-caching --override-generation-config '{"temperature":0.7,"top_p":0.95}'
localmaxxing · run id
cmrxk8q0x00dco401yhwzjx9m
localmaxxing · tokenized · arguments
docker, run, --gpus, all, --ipc=host, --network=host, --shm-size=32g, -e, CUTE_DSL_ARCH=sm_120a, -v, <MODEL_DIR>:/model:ro, vllm/vllm-openai:<nightly-cu13>, --model, /model, --served-model-name, laguna, --trust-remote-code, --tensor-parallel-size, 4, --disable-custom-all-reduce, --dtype, bfloat16, --kv-cache-dtype, fp8, --max-model-len, 262144, --gpu-memory-utilization, 0.90, --max-num-seqs, 1, --max-num-batched-tokens, 8192, --enable-prefix-caching, --override-generation-config, {"temperature":0.7,"top_p":0.95}
localmaxxing · tokenized · fidelity
faithful

provenance

captured at
2026-08-30T09:10:02Z

sources

captured atkindurl
2026-08-30T09:10:02Znormalized-recipewww.localmaxxing.com/en/runs/cmrxk8q0x00dco401yhwzjx9m ↗