Recipe

qwen3-6-27b-fp8-dgx-spark-gb10-128gb-sglang-tp1

qwen3-6-27b-fp8-dgx-spark-gb10-128gb-sglang-tp1

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

Record

Status
candidate
Source
localmaxxing
Engine
sglang
Engine version
0.5.9
Accelerators
1
Tensor parallel
1
Context tokens
65,536
Max concurrency
1
chat
unknown
reasoning
unknown
tools
unknown
vision
unknown

Hugging Face model card

Identity

https://huggingface.co/Qwen/Qwen3.6-27B-FP8
Repository
Qwen/Qwen3.6-27B-FP8
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

sglang

Evidence only · candidate · reference

Candidate evidence — not a Run contract

Source
https://www.localmaxxing.com/en/runs/cmof6aro10002l404x6vocua9

Observed source tokens

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

  1. python
  2. -m
  3. sglang.launch_server
  4. --model-path
  5. /home/radar/models/Qwen3.6-27B-FP8
  6. --fp8-gemm-backend
  7. triton
  8. --attention-backend
  9. triton
  10. --context-length
  11. 65536
FlagValue
-msglang.launch_server
--model-path/home/radar/models/Qwen3.6-27B-FP8
--fp8-gemm-backendtriton
--attention-backendtriton
--context-length65536

Measured speed

ConcurrencyContextPrefillDecodeTTFT msStatusSweep
165,5364.7observedqwen3-6-27b-fp8-dgx-spark-gb10-128gb-sglang-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
qwen3-6-27b-fp8-dgx-spark-gb10-128gb-sglang-tp1
model instance id
qwen-qwen3-6-27b-fp8--fp8
recipe source
localmaxxing
schema version
local-ai-registry/v1
speed sweep ids
qwen3-6-27b-fp8-dgx-spark-gb10-128gb-sglang-tp1-sweep
status
candidate

capabilities

draft launch

accelerator backend
nvidia
arguments
python3, -m, sglang.launch_server, --model-path, Qwen/Qwen3.6-27B-FP8, --tp, 1, --host, 0.0.0.0, --port, 30000, --context-length, 65536
container port
30,000
host port
30,000
image
lmsysorg/sglang:dev-cu13@sha256:6cd4635214f279e0a43019f88e3120d407567640a58aa7dcc0085e3d91402cc4
ipc
host
kind
docker
shm size
16g
synthesized · generated at
2026-08-31T22:12:17Z
synthesized · image provenance
gemma-4-12b-it-nvfp4-rtxpro6000-sglang-tp1
synthesized · template
sglang-launch-server-v1

mounts

read onlytarget
No/root/.cache/huggingface

engine

name
sglang
version
0.5.9

serving

max concurrency
1
max context tokens
65,536
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
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

metadata

localmaxxing · backend
cuda
localmaxxing · hardware label
GB10 Grace Blackwell
localmaxxing · notes
DGX Spark SGLang FP8 triton GEMM path for Qwen3.6-27B-FP8; compatible but much slower than vLLM MTP2. Source: /home/radar/qwen36_27b_fp8_vllm_mtp_runfile_2026-04-23.md
localmaxxing · observed command
python -m sglang.launch_server --model-path /home/radar/models/Qwen3.6-27B-FP8 --fp8-gemm-backend triton --attention-backend triton --context-length 65536
localmaxxing · run id
cmof6aro10002l404x6vocua9
localmaxxing · tokenized · arguments
python, -m, sglang.launch_server, --model-path, /home/radar/models/Qwen3.6-27B-FP8, --fp8-gemm-backend, triton, --attention-backend, triton, --context-length, 65536
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/cmof6aro10002l404x6vocua9