Speed test matrix

GLM-5.3 on H200 SXM 141GB ×8

vllm · 0.29.0 · FP86 contexts (8K–251K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 64K, 128K, 251K, 256K

by @keennaylmx dev importMeasured Sep 14, 2026, 19:29 UTC

Hardware

Type: Discrete GPU
GPU: H200 SXM
GPU count: 8
VRAM: 1128 GB (8 × 141 GB)
OS: linux

Engine

Engine: vllm
Version: 0.29.0
Quantization: FP8
Backend: cuda
Repository: github.com/vllm-project/vllm
Build: cuda-13.2

Run

Served model: glm
Revision: main
Benchmark tool: llm-inference-bench v0.6.2
Decode test: Fixed duration
Prefill mode: Measured during decode warm-up
Time per measurement: 30 s
Max output: 4,096 tokens

Launch command

"$HOME/env_h200-glm53-vllm-v0290/bin/vllm" serve zai-org/GLM-5.3 --served-model-name glm --tensor-parallel-size 8 --reasoning-parser glm45 --enable-auto-tool-choice --tool-call-parser glm47 --max-model-len 262144 --gpu-memory-utilization 0.87 --host 0.0.0.0 --port 8000 --api-key <redacted> --dtype bfloat16 --kv-cache-dtype bfloat16 --no-enable-flashinfer-autotune --speculative-config '{"method":"mtp","num_speculative_tokens":5}'

Prefill

prompt tok/s by context length

8K context11.9K688 ms TTFT

16K context11.7K1,391 ms TTFT

32K context11.3K2,859 ms TTFT

64K context10.6K6,109 ms TTFT

128K context9,46113,623 ms TTFT

251K context7,71132,757 ms TTFT

256K context7,70433,443 ms TTFT

Sustained decode

total tok/s · rows are context length, columns are simultaneous requests

Context	1	2	4	8	16	32	64	128
8K	170.963 ms TTFT	302.788 ms TTFT	484.3101 ms TTFT	722.5164 ms TTFT	1,203.0290 ms TTFT	Over capacity	Over capacity	Over capacity
16K	164.184 ms TTFT	319.7103 ms TTFT	472.4110 ms TTFT	704.0241 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity
32K	186.0122 ms TTFT	307.2143 ms TTFT	494.2158 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
64K	181.4205 ms TTFT	302.1213 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
128K	167.3384 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
251K	172.2809 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity

Over capacity: not run because this many requests at this context would exceed the server's KV cache.

TTFT is measured after the benchmark warms the server's prompt cache, so it mostly excludes prefill.

All measurements

Latencies are medians. Total tok/s is output across all simultaneous requests.

Phase	Context	Requests at once	Status	Total tok/s	Per request tok/s	TTFT	Time between tokens	Average active requests	Completed requests	Output tokens	Measured for
Prefill	8K	1	Valid	11.9K	—	688 ms	—	—	1	—	—
Prefill	16K	1	Valid	11.7K	—	1,391 ms	—	—	1	—	—
Prefill	32K	1	Valid	11.3K	—	2,859 ms	—	—	1	—	—
Prefill	64K	1	Valid	10.6K	—	6,109 ms	—	—	1	—	—
Prefill	128K	1	Valid	9,461.0	—	13,623 ms	—	—	1	—	—
Prefill	251K	1	Valid	7,711.0	—	32,757 ms	—	—	1	—	—
Prefill	256K	1	Valid	7,704.0	—	33,443 ms	—	—	1	—	—
Sustained decode	8K	1	Valid	170.9	174.7	63 ms	5.7 ms	1.0	1	5,126	30.0 s
Sustained decode	8K	2	Valid	302.7	156.1	88 ms	6.4 ms	2.0	2	9,076	30.0 s
Sustained decode	8K	4	Valid	484.3	128.4	101 ms	7.8 ms	4.0	3	14,522	30.0 s
Sustained decode	8K	8	Valid	722.5	89.3	164 ms	11.2 ms	8.0	0	21,676	30.0 s
Sustained decode	8K	16	Valid	1,203.0	75.8	290 ms	13.2 ms	16.0	0	36,078	30.0 s
Sustained decode	8K	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	8K	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	8K	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	1	Valid	164.1	169.3	84 ms	5.9 ms	1.0	1	4,921	30.0 s
Sustained decode	16K	2	Valid	319.7	163.9	103 ms	6.1 ms	2.0	2	9,588	30.0 s
Sustained decode	16K	4	Valid	472.4	129.6	110 ms	7.7 ms	4.0	3	14,169	30.0 s
Sustained decode	16K	8	Valid	704.0	90.0	241 ms	11.1 ms	8.0	0	21,099	30.0 s
Sustained decode	16K	16	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	32K	1	Valid	186.0	185.4	122 ms	5.4 ms	1.0	1	5,579	30.0 s
Sustained decode	32K	2	Valid	307.2	155.4	143 ms	6.4 ms	2.0	2	9,213	30.0 s
Sustained decode	32K	4	Valid	494.2	132.7	158 ms	7.5 ms	4.0	4	14,826	30.0 s
Sustained decode	32K	8	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	32K	16	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	32K	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	32K	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	32K	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	64K	1	Valid	181.4	180.8	205 ms	5.6 ms	1.0	1	5,441	30.0 s
Sustained decode	64K	2	Valid	302.1	154.7	213 ms	6.5 ms	2.0	2	9,059	30.0 s
Sustained decode	64K	4	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	64K	8	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	64K	16	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	64K	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	64K	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	64K	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	128K	1	Valid	167.3	170.6	384 ms	5.9 ms	1.0	1	5,016	30.0 s
Sustained decode	128K	2	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	128K	4	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	128K	8	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	128K	16	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	128K	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	128K	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	128K	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	251K	1	Valid	172.2	178.6	809 ms	5.6 ms	1.0	1	5,166	30.0 s
Sustained decode	251K	2	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	251K	4	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	251K	8	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	251K	16	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	251K	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	251K	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	251K	128	Over capacity	—	—	—	—	—	—	—	—

Over capacity: not run because this many requests at this context would exceed the server's KV cache.