LLM Inference Benchmark Explorer
Qwen3.6-27B on 2× DGX Spark — NVFP4, vLLM, TP2 inference benchmark
Open this configuration in the LLM Inference Benchmark Explorer
| Model | Qwen3.6-27B |
|---|---|
| Parameters | 27B |
| Intelligence Index | 21.4 |
| Agentic Index | 18.5 |
| Device | 2× DGX Spark |
| Quantization | NVFP4 |
| Max C | 4 |
| Chat Capacity | 16 |
| Agentic Capacity | 6 |
| TP | 2 |
| DP | — |
| PP | — |
| Engine | vLLM |
| Speculative Decoding | — |
| C | TTFT (ms) | TPS (tok/s) | Status |
|---|---|---|---|
| 1 | 149 | 23.00 | PASS |
| 2 | 252 | 21.72 | PASS |
| 4 | 320 | 20.77 | PASS |
| 8 | 457 | 18.70 | FAIL |
| 16 | 741 | 16.25 | FAIL |
| 32 | 1197 | 12.24 | FAIL |
| 64 | 2137 | 7.58 | FAIL |
In this measurement, Qwen3.6-27B (27B parameters), served in NVFP4 format with vLLM on 2× DGX Spark (TP=2), reached a generation speed of 23 tok/s per request and a time to first token (TTFT) of 149 ms with a single request.
Considering the speed targets and the available KV cache capacity, the estimated capacity is 16 users for chat use and 6 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.
In this measurement, Qwen3.6-27B (27B parameters), served in NVFP4 format with vLLM on 2× DGX Spark (TP=2), reached a generation speed of 23 tok/s per request and a time to first token (TTFT) of 149 ms with a single request.
Considering the speed targets and the available KV cache capacity, the estimated capacity is 16 users for chat use and 6 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.