Local AI Registry

Recipe

qwen3-6-27b-nvfp4-dgx-spark-gb10-128gb-vllm-tp1-sg1v1o0x

qwen3-6-27b-nvfp4-dgx-spark-gb10-128gb-vllm-tp1-sg1v1o0x
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
v0.20.2rc1.dev166+gf6490a284 / ghcr.io/aeon-7/vllm-aeon-ultimate-dflash:qwen36-v4
Accelerators
1
Tensor parallel
1
Context tokens
262,144
Max concurrency
64
chat
unknown
reasoning
unknown
tools
unknown
vision
unknown

Hugging Face model card

Identity

https://huggingface.co/Qwen/Qwen3.6-27B
Repository
Qwen/Qwen3.6-27B
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/cmp8te6o3035mo401sg1v1o0x

Observed source tokens

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

  1. vllm
  2. serve
  3. /models/aeon-xs
  4. --served-model-name
  5. aeon-ultimate
  6. qwen36-ultimate
  7. aeon-fast
  8. aeon-deep
  9. qwen36-ultimate-xs
  10. --host
  11. 0.0.0.0
  12. --port
  13. 8000
  14. --tensor-parallel-size
  15. 1
  16. --dtype
  17. auto
  18. --quantization
  19. modelopt
  20. --kv-cache-dtype
  21. auto
  22. --max-model-len
  23. 262144
  24. --max-num-seqs
  25. 64
  26. --max-num-batched-tokens
  27. 32768
  28. --gpu-memory-utilization
  29. 0.85
  30. --enable-chunked-prefill
  31. --no-enable-prefix-caching
  32. --generation-config
  33. vllm
  34. --load-format
  35. safetensors
  36. --trust-remote-code
  37. --enable-auto-tool-choice
  38. --tool-call-parser
  39. qwen3_coder
  40. --reasoning-parser
  41. qwen3
  42. --attention-backend
  43. flash_attn
  44. --compilation-config
  45. \{\"inductor_compile_config\":\{\"combo_kernels\":false\,\"benchmark_combo_kernel\":false\}\}
  46. --limit-mm-per-prompt
  47. \{\"image\":4\,\"video\":2\}
  48. --mm-encoder-tp-mode
  49. data
  50. --mm-processor-cache-type
  51. shm
  52. --mm-shm-cache-max-object-size-mb
  53. 256
  54. --speculative-config
  55. \{\"method\":\"dflash\"\,\"model\":\"/models/dflash-drafter\"\,\"num_speculative_tokens\":15\,\"attention_backend\":\"FLASH_ATTN\"\}
FlagValue
--served-model-nameaeon-ultimate
--host0.0.0.0
--port8000
--tensor-parallel-size1
--dtypeauto
--quantizationmodelopt
--kv-cache-dtypeauto
--max-model-len262144
--max-num-seqs64
--max-num-batched-tokens32768
--gpu-memory-utilization0.85
--generation-configvllm
--load-formatsafetensors
--tool-call-parserqwen3_coder
--reasoning-parserqwen3
--attention-backendflash_attn
--compilation-config\{\"inductor_compile_config\":\{\"combo_kernels\":false\,\"benchmark_combo_kernel\":false\}\}
--limit-mm-per-prompt\{\"image\":4\,\"video\":2\}
--mm-encoder-tp-modedata
--mm-processor-cache-typeshm
--mm-shm-cache-max-object-size-mb256
--speculative-config\{\"method\":\"dflash\"\,\"model\":\"/models/dflash-drafter\"\,\"num_speculative_tokens\":15\,\"attention_backend\":\"FLASH_ATTN\"\}

Measured speed

ConcurrencyContextPrefillDecodeTTFT msStatusSweep
1262,1442,286.134.9900.2observedqwen3-6-27b-nvfp4-dgx-spark-gb10-128gb-vllm-tp1-sg1v1o0x-sweep

Remaining fields

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

hardware count
1
hardware id
dgx-spark-gb10-128gb
id
qwen3-6-27b-nvfp4-dgx-spark-gb10-128gb-vllm-tp1-sg1v1o0x
model instance id
qwen-qwen3-6-27b--nvfp4
recipe source
localmaxxing
schema version
local-ai-registry/v1
speed sweep ids
qwen3-6-27b-nvfp4-dgx-spark-gb10-128gb-vllm-tp1-sg1v1o0x-sweep
status
candidate

capabilities

draft launch

accelerator backend
nvidia
arguments
serve, --model, Qwen/Qwen3.6-27B, --tensor-parallel-size, 1, --host, 0.0.0.0, --port, 8000, --max-model-len, 262144
container port
8,000
entrypoint
vllm
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
v0.20.2rc1.dev166+gf6490a284 / ghcr.io/aeon-7/vllm-aeon-ultimate-dflash:qwen36-v4

serving

max concurrency
64
max context tokens
262,144
tensor parallel
1
Provenance & metadata (3)

facts

capabilities.chat · provenance · captured at
2026-08-30T09:26:09Z
capabilities.chat · reason capability-not-verified
capabilities.chat · state unknown
capabilities.reasoning · provenance · captured at
2026-08-30T09:26:09Z
capabilities.reasoning · reason capability-not-verified
capabilities.reasoning · state unknown
capabilities.tools · provenance · captured at
2026-08-30T09:26:09Z
capabilities.tools · reason capability-not-verified
capabilities.tools · state unknown
capabilities.vision · provenance · captured at
2026-08-30T09:26:09Z
capabilities.vision · reason capability-not-verified
capabilities.vision · state unknown
draft launch.entrypoint · provenance · captured at
2026-09-01T01:41:26Z
draft launch.entrypoint · reason linux-amd64-container-config-entrypoint
draft launch.entrypoint · state known
engine.graph mode · provenance · captured at
2026-08-30T09:26:09Z
engine.graph mode · reason runtime-detail-not-published
engine.graph mode · state unknown
serving.kv cache tokens · provenance · captured at
2026-08-30T09:26:09Z
serving.kv cache tokens · reason kv-cache-capacity-not-published
serving.kv cache tokens · state unknown
serving.max concurrency · provenance · captured at
2026-08-31T23:03:15Z
serving.max concurrency · reason server-capacity-derived-from-source-evidence
serving.max concurrency · state known

metadata

localmaxxing · backend
cuda
localmaxxing · batch size
1
localmaxxing · hardware label
GB10 Grace Blackwell
localmaxxing · notes
Submitted under parent/root hfId Qwen/Qwen3.6-27B for leaderboard grouping; actual served weights are AEON-7/Qwen3.6-27B-AEON-Ultimate-Uncensored-Multimodal-NVFP4-MTP-XS (ModelOpt NVFP4) from /models/aeon-xs. Cell=dflash/think-on/code, spark-3 DGX Spark, vLLM/FlashAttention, llama-benchy 0.3.7, pp=2048 tg=128 c=1 runs=30, first sample dropped (warm n=29). tokSOut=34.94, tokSPrefill=2286.10, tokSTotal=477.29, ttftMs=900.22. DFlash uses z-lab/Qwen3.6-27B-DFlash with num_speculative_tokens=15; companion AR row has specDecoding=false. Raw receipts + submission package: https://github.com/my-other-github-account/spark-bench-reproducers/tree/main/qwen36-aeon-combined-flashqla-dflash-spark/submission-tg128-pp2048-balanced-root-hfid-20260516 and https://github.com/my-other-github-account/spark-bench-reproducers/tree/main/qwen36-aeon-combined-flashqla-dflash-spark/results/tg128-pp2048-code-prose-think-grid-20260516_054837. Commit 47d3c4b. By @banana_baeee.
localmaxxing · observed command
vllm serve /models/aeon-xs --served-model-name aeon-ultimate qwen36-ultimate aeon-fast aeon-deep qwen36-ultimate-xs --host 0.0.0.0 --port 8000 --tensor-parallel-size 1 --dtype auto --quantization modelopt --kv-cache-dtype auto --max-model-len 262144 --max-num-seqs 64 --max-num-batched-tokens 32768 --gpu-memory-utilization 0.85 --enable-chunked-prefill --no-enable-prefix-caching --generation-config vllm --load-format safetensors --trust-remote-code --enable-auto-tool-choice --tool-call-parser qwen3_coder --reasoning-parser qwen3 --attention-backend flash_attn --compilation-config \{\"inductor_compile_config\":\{\"combo_kernels\":false\,\"benchmark_combo_kernel\":false\}\} --limit-mm-per-prompt \{\"image\":4\,\"video\":2\} --mm-encoder-tp-mode data --mm-processor-cache-type shm --mm-shm-cache-max-object-size-mb 256 --speculative-config \{\"method\":\"dflash\"\,\"model\":\"/models/dflash-drafter\"\,\"num_speculative_tokens\":15\,\"attention_backend\":\"FLASH_ATTN\"\}
localmaxxing · run id
cmp8te6o3035mo401sg1v1o0x
localmaxxing · tokenized · arguments
vllm, serve, /models/aeon-xs, --served-model-name, aeon-ultimate, qwen36-ultimate, aeon-fast, aeon-deep, qwen36-ultimate-xs, --host, 0.0.0.0, --port, 8000, --tensor-parallel-size, 1, --dtype, auto, --quantization, modelopt, --kv-cache-dtype, auto, --max-model-len, 262144, --max-num-seqs, 64, --max-num-batched-tokens, 32768, --gpu-memory-utilization, 0.85, --enable-chunked-prefill, --no-enable-prefix-caching, --generation-config, vllm, --load-format, safetensors, --trust-remote-code, --enable-auto-tool-choice, --tool-call-parser, qwen3_coder, --reasoning-parser, qwen3, --attention-backend, flash_attn, --compilation-config, \{\"inductor_compile_config\":\{\"combo_kernels\":false\,\"benchmark_combo_kernel\":false\}\}, --limit-mm-per-prompt, \{\"image\":4\,\"video\":2\}, --mm-encoder-tp-mode, data, --mm-processor-cache-type, shm, --mm-shm-cache-max-object-size-mb, 256, --speculative-config, \{\"method\":\"dflash\"\,\"model\":\"/models/dflash-drafter\"\,\"num_speculative_tokens\":15\,\"attention_backend\":\"FLASH_ATTN\"\}
localmaxxing · tokenized · fidelity
faithful

provenance

captured at
2026-08-30T09:26:09Z

sources

captured atkindurl
2026-08-30T09:26:09Znormalized-recipewww.localmaxxing.com/en/runs/cmp8te6o3035mo401sg1v1o0x ↗