Qwen3.6-27B · SGLang · NVFP4

status: done

Configuration

ModelQwen/Qwen3.6-27B
CompanyAlibaba
FamilyQwen
Parameters27B (dense)
EngineSGLang
Quant / precisionNVFP4
Why this quantUnsloth's NVFP4 (W4A4) quant of Qwen3.6-27B (unsloth/Qwen3.6-27B-NVFP4) on SGLang — the card lists SGLang as a supported path. Base (non-speculative). Required a NEWER SGLang nightly (nightly-dev-cu13-20260623, transformers 5.8.1) — the stock spark image can't load the qwen3.6 arch. Cross-engine compare vs the vLLM NVFP4 run.
Downloadunsloth/Qwen3.6-27B-NVFP4
Context window65536
Input modalitiestext, image, video (served text-only here)

Measured results

Prefill tok/s179.07
Decode tok/s177.99
Peak memory (GB)103.21 (system MemAvailable delta (10s sampling) — SGLang static KV reservation; weights 24.55 GB resident (from log))
Completed2026-06-23 11:35 +08

Full run command

# Stock spark image CAN'T load qwen3.6 (transformers 4.57.1). Used a newer SGLang nightly
# (nightly-dev-cu13-20260623-ba9d5aed, arm64, transformers 5.8.1) via the new SGLANG_IMAGE override.
SGLANG_IMAGE=lmsysorg/sglang:nightly-dev-cu13-20260623-ba9d5aed \
  scripts/bench-sglang-serving.sh unsloth/Qwen3.6-27B-NVFP4 65536 32 1000 900 256 --trust-remote-code
# 647/1000 prompts, 0 errors (hit 900 s cap). ready after 276 s. TTFT median 2169 ms, TPOT median 159 ms.
# Load log: type=Qwen3_5ForConditionalGeneration, quant=compressed-tensors, weights 24.55 GB.

qwen3.6-27b Alibaba Qwen NVFP4 16-40B conc-32

Notes

UNBLOCKED on a newer SGLang nightly — cross-engine NVFP4 datapoint vs vLLM. The stock lmsysorg/sglang:spark image (transformers 4.57.1) couldn’t load the Qwen3.6 (qwen3_5) arch; the nightly-dev-cu13-20260623-ba9d5aed image (arm64, transformers 5.8.1) loads it fine.

  • Result (conc 32): prefill 179.1 / decode 177.99 tok/s aggregate; 647/1000, 0 errors (hit the 900 s cap). Peak mem 103.2 GB; weights 24.55 GB resident (log) — confirms NVFP4 is active (a BF16 fallback would be ~54 GB and far slower).
  • Cross-engine vs vLLM: decode 178.0 (SGLang) vs 187.7 (vLLM) on the same unsloth NVFP4 / 65536 / conc-32 — vLLM ~5% faster on decode, SGLang ~5 GB leaner on peak mem. Close; neither dominates for this 27B NVFP4 base.
  • NVFP4 caveat (benign): SGLang logs “Acceleration for non-quantized schemes is not supported by Compressed Tensors. Falling back to UnquantizedLinearMethod” — this applies only to the model’s non-quantized layers (norms/embeddings); quant=compressed-tensors + the 24.55 GB footprint + the 178 tok/s decode all confirm the 4-bit expert/linear weights are used.
  • Tokenizer warning: “Tokenizer … is still TokenizersBackend … attributes may be missing” — did not cause errors (0 failed requests), noted in case it matters for a later run.
  • Engine image: pinned the nightly via the new SGLANG_IMAGE= wrapper override. The blocked SGLang Qwen3.6 siblings (27B MTP, 35B-A3B base/MTP) are unblocked the same way.