LLM Inference Benchmark Explorer

GLM-5.3-Flash on DGX B300 — FP8, SGLang, TP4 inference benchmark

Open this configuration in the LLM Inference Benchmark Explorer

ModelGLM-5.3-Flash
Parameters321B
Intelligence Index41.8
Agentic Index50.9
DeviceDGX B300
QuantizationFP8
Max C32
Chat Capacity128
Agentic Capacity48
TP4
DP—
PP—
EngineSGLang
Speculative Decoding—
CTTFT (ms)TPS (tok/s)Status
1292245.10PASS
2323157.48PASS
4338107.07PASS
845857.02PASS
16296636.38FAIL
3254232.74PASS
64201225.66FAIL

In this measurement, GLM-5.3-Flash (321B parameters), served in FP8 format with SGLang on DGX B300 (TP=4), reached a generation speed of 245.1 tok/s per request and a time to first token (TTFT) of 292 ms with a single request.

Considering the speed targets and the available KV cache capacity, the estimated capacity is 128 users for chat use and 48 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.