Recipe

qwen3-6-35b-a3b-heretic-nvfp4-compressed-tensors-dgx-spark-gb10-128gb-vllm-tp1

qwen3-6-35b-a3b-heretic-nvfp4-compressed-tensors-dgx-spark-gb10-128gb-vllm-tp1

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

Record

Status
candidate
Source
localmaxxing
Engine
vllm
Engine version
0.1.dev1+gbfde49e28.d20260418
Accelerators
1
Tensor parallel
1
Context tokens
262,144
Max concurrency
128
chat
unknown
reasoning
unknown
tools
unknown
vision
unknown

Hugging Face model card

Identity

https://huggingface.co/AEON-7/Qwen3.6-35B-A3B-heretic-NVFP4
Repository
AEON-7/Qwen3.6-35B-A3B-heretic-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/cmozhdij2001elo014mcecsr3

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/qwen36
  4. --tensor-parallel-size
  5. 1
  6. --quantization
  7. compressed-tensors
  8. --max-model-len
  9. 262144
  10. --max-num-seqs
  11. 128
  12. --max-num-batched-tokens
  13. 65536
  14. --gpu-memory-utilization
  15. 0.85
  16. --enable-chunked-prefill
  17. --enable-prefix-caching
  18. --speculative-config
  19. {method:dflash,model:/models/qwen36-dflash,num_speculative_tokens:15}
  20. --tool-call-parser
  21. qwen3_coder
  22. --reasoning-parser
  23. qwen3
  24. --attention-backend
  25. flash_attn
FlagValue
--tensor-parallel-size1
--quantizationcompressed-tensors
--max-model-len262144
--max-num-seqs128
--max-num-batched-tokens65536
--gpu-memory-utilization0.85
--speculative-config{method:dflash,model:/models/qwen36-dflash,num_speculative_tokens:15}
--tool-call-parserqwen3_coder
--reasoning-parserqwen3
--attention-backendflash_attn

Measured speed

ConcurrencyContextPrefillDecodeTTFT msStatusSweep
1262,14478.9135.2observedqwen3-6-35b-a3b-heretic-nvfp4-compressed-tensors-dgx-spark-gb10-128gb-vllm-tp1-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
qwen3-6-35b-a3b-heretic-nvfp4-compressed-tensors-dgx-spark-gb10-128gb-vllm-tp1
model instance id
aeon-7-qwen3-6-35b-a3b-heretic-nvfp4--nvfp4-compressed-tensors
recipe source
localmaxxing
schema version
local-ai-registry/v1
speed sweep ids
qwen3-6-35b-a3b-heretic-nvfp4-compressed-tensors-dgx-spark-gb10-128gb-vllm-tp1-sweep
status
candidate

capabilities

draft launch

accelerator backend
nvidia
arguments
serve, --model, AEON-7/Qwen3.6-35B-A3B-heretic-NVFP4, --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
0.1.dev1+gbfde49e28.d20260418

serving

max concurrency
128
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 · batch size
1
localmaxxing · hardware label
GB10 Grace Blackwell
localmaxxing · notes
AEON-7 Qwen3.6-35B-A3B-heretic NVFP4 + DFlash spec decoding (k=15) on SINGLE DGX Spark. MoE with ~3B active params. tokSOut=78.9 t/s median c=1 (5 runs max_tokens=256, mean 81.98, stdev 9.71, max 96.37). TTFR 135ms median. KV cache 601520 tokens = 4.86x concurrency at 262K context. Image: ghcr.io/aeon-7/vllm-spark-omni-q36:v1.2. Free Spark 1 to AEON-7 27B XS - dual-Spark dual-model split deploy. Synced from Spark 1 to Spark 2 over QSFP56 200G.
localmaxxing · observed command
vllm serve /models/qwen36 --tensor-parallel-size 1 --quantization compressed-tensors --max-model-len 262144 --max-num-seqs 128 --max-num-batched-tokens 65536 --gpu-memory-utilization 0.85 --enable-chunked-prefill --enable-prefix-caching --speculative-config {method:dflash,model:/models/qwen36-dflash,num_speculative_tokens:15} --tool-call-parser qwen3_coder --reasoning-parser qwen3 --attention-backend flash_attn
localmaxxing · run id
cmozhdij2001elo014mcecsr3
localmaxxing · tokenized · arguments
vllm, serve, /models/qwen36, --tensor-parallel-size, 1, --quantization, compressed-tensors, --max-model-len, 262144, --max-num-seqs, 128, --max-num-batched-tokens, 65536, --gpu-memory-utilization, 0.85, --enable-chunked-prefill, --enable-prefix-caching, --speculative-config, {method:dflash,model:/models/qwen36-dflash,num_speculative_tokens:15}, --tool-call-parser, qwen3_coder, --reasoning-parser, qwen3, --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/cmozhdij2001elo014mcecsr3