benchmarks / qwen3-6-35b-a3b--dgx-spark-x1--vllm--bf16--p2048-o256-c1--2026-07-06
Qwen3.6 35B-A3B on NVIDIA DGX Spark
compare30.7tok/s
544msttft
3.7kprefill
vLLM 0.24.0BF16Jul 6, 2026
Run configuration
nodes
1
batch size
—
concurrency
1
context
10240
prompt / output
2041 / 256
weights Qwen/Qwen3.6-35B-A3B ↗
Metrics
decode tok/s
30.7
prefill tok/s
3.7k
ttft
544ms
total tok/s
28.9
inter-token
33ms
e2e latency
8.86s
req/s
0.113
peak memory
83.9 GB
avg power
28.8 W
energy/req
255.3 J
Reproduce
PATH="$HOME/venvs/vllm/bin:$PATH" VLLM_USE_DEEP_GEMM=0 TORCHINDUCTOR_COMPILE_THREADS=2 MAX_JOBS=4 vllm serve ~/models/hf/Qwen3.6-35B-A3B --served-model-name Qwen/Qwen3.6-35B-A3B --max-model-len 10240 --gpu-memory-utilization 0.62 --max-num-seqs 8 --limit-mm-per-prompt '{"image": 0, "video": 0}' --port 8000
python3 bench/harness.py bench/configs/qwen3-6-35b-a3b-bf16--vllm.json --rows-out data/benchmarks/qwen3-6-35b-a3b.jsonUnquantized baseline (original safetensors), vLLM on GB10