LLM Inference Benchmark Explorer

DeepSeek-V3.1 on DGX B300 — NVFP4, vLLM, TP2 inference benchmark

Open this configuration in the LLM Inference Benchmark Explorer

ModelDeepSeek-V3.1
Parameters685B
Intelligence Index13.7
Agentic Index—
DeviceDGX B300
QuantizationNVFP4
Max C32+
Chat Capacity68
Agentic Capacity17
TP2
DP—
PP—
EnginevLLM
Speculative Decoding—
CTTFT (ms)TPS (tok/s)Status
1140122.25PASS
2291113.13PASS
4288105.21PASS
828796.15PASS
1629984.38PASS
3235374.36PASS

In this measurement, DeepSeek-V3.1 (685B parameters), served in NVFP4 format with vLLM on DGX B300 (TP=2), reached a generation speed of 122.2 tok/s per request and a time to first token (TTFT) of 140 ms with a single request.

Considering the speed targets and the available KV cache capacity, the estimated capacity is 68 users for chat use and 17 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.

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