LLM Inference Benchmark Explorer
GPT-OSS 120B on 1× DGX Spark — MXFP4, vLLM, TP1 inference benchmark
Open this configuration in the LLM Inference Benchmark Explorer
| Model | GPT-OSS 120B |
|---|---|
| Parameters | 120B |
| Intelligence Index | 11.6 |
| Agentic Index | 3.7 |
| Device | 1× DGX Spark |
| Quantization | MXFP4 |
| Max C | 4 |
| Chat Capacity | 16 |
| Agentic Capacity | 6 |
| TP | — |
| DP | — |
| PP | — |
| Engine | vLLM |
| Speculative Decoding | — |
| C | TTFT (ms) | TPS (tok/s) | Status |
|---|---|---|---|
| 1 | 220 | 61.09 | PASS |
| 2 | 294 | 40.19 | PASS |
| 4 | 321 | 26.70 | PASS |
| 8 | 395 | 17.66 | FAIL |
| 16 | 444 | 12.08 | FAIL |
| 32 | 552 | 8.51 | FAIL |
In this measurement, GPT-OSS 120B (120B parameters), served in MXFP4 format with vLLM on 1× DGX Spark, reached a generation speed of 61.1 tok/s per request and a time to first token (TTFT) of 220 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: CUTLASS MoE, FlashInfer attn, GMU 0.7, KV fp8, Ray
Intelligence Index and Agentic Index values are published by Artificial Analysis and are reproduced here with attribution.
In this measurement, GPT-OSS 120B (120B parameters), served in MXFP4 format with vLLM on 1× DGX Spark, reached a generation speed of 61.1 tok/s per request and a time to first token (TTFT) of 220 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.