LLM Inference Benchmark Explorer

Qwen3.6-27B on 4× DGX Spark — NVFP4, vLLM, TP4 inference benchmark

Open this configuration in the LLM Inference Benchmark Explorer

ModelQwen3.6-27B
Parameters27B
Intelligence Index21.4
Agentic Index18.5
Device4× DGX Spark
QuantizationNVFP4
Max C8
Chat Capacity32
Agentic Capacity12
TP4
DP—
PP—
EnginevLLM
Speculative Decoding—
CTTFT (ms)TPS (tok/s)Status
116433.11PASS
226729.93PASS
437327.51PASS
844624.98PASS
1680519.21FAIL
32100514.61FAIL
6415517.93FAIL

In OpenZeka's measurement, Qwen3.6-27B (27B parameters), served in NVFP4 format with vLLM on 4× DGX Spark (TP=4), reached a generation speed of 33.1 tok/s per request and a time to first token (TTFT) of 164 ms with a single request.

Considering the speed targets and the available KV cache capacity, the estimated capacity is 32 users for chat use and 12 for agentic use. The realistic capacity will likely fall between these two values. In scenarios dominated by coding, tool use, long workflows and multi-agent use, capacity approaches the agentic estimate; where shorter interactions, standard conversations and lighter tasks dominate, it approaches the chat estimate.

Notes: nvidia NVFP4, eugr nightly

Intelligence Index and Agentic Index values are published by Artificial Analysis and are reproduced here with attribution.