LLM Inference Benchmark Explorer

Nemotron-3-Ultra on DGX B300 — BF16, vLLM, TP8 inference benchmark

Open this configuration in the LLM Inference Benchmark Explorer

ModelNemotron-3-Ultra
Parameters550B
Intelligence Index22.9
Agentic Index20.1
DeviceDGX B300
QuantizationBF16
Max C16
Chat Capacity64
Agentic Capacity24
TP8
DP—
PP—
EnginevLLM
Speculative Decoding—
CTTFT (ms)TPS (tok/s)Status
1132204.92PASS
2465128.50PASS
4134131.41PASS
8159106.38PASS
1626660.31PASS
32215966.84FAIL
64606466.93FAIL

In this measurement, Nemotron-3-Ultra (550B parameters), served in BF16 format with vLLM on DGX B300 (TP=8), reached a generation speed of 204.9 tok/s per request and a time to first token (TTFT) of 132 ms with a single request.

Considering the speed targets and the available KV cache capacity, the estimated capacity is 64 users for chat use and 24 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.