Speed test matrix

Qwen3.5-4B-FP8-dynamic on RTX 3090 24GB

vllm · 0.32.0 · FP8 dynamic6 contexts (0–248K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 64K, 128K

by @Sandbergllm-inference-bench v0.8.0Measured Oct 30, 2026, 07:11 UTC

Hardware

Type: Discrete GPU
GPU: RTX 3090
VRAM: 24 GB
CPU: Intel Core i7-9700K (8C)
RAM: 32 GB
OS: Ubuntu 24.04.5

Engine

Engine: vllm
Version: 0.32.0
Quantization: FP8 dynamic
Backend: cuda
Repository: github.com/vllm-project/vllm
Build: vllm/vllm-openai:v0.32.0-x86_64

Run

Served model: bench
Revision: main
Benchmark tool: llm-inference-bench v0.8.0
Decode test: Fixed duration
Prefill mode: Measured during decode warm-up
Time per measurement: 30 s
Max output: 8,192 tokens

Launch command

# Untuned out-of-the-box baseline. Spec decode: none
# 1x RTX 3090 24GB (Zagitta/gpu-unlock), 420 W limit, PCIe Gen3 x16 | Intel Core i7-9700K 8C | 32 GB DDR4 | Ubuntu 24.04.5, NVIDIA driver 580.178.04
vllm serve RedHatAI/Qwen3.5-4B-FP8-dynamic --max-model-len 262144 --gpu-memory-utilization 0.94 --max-num-seqs 128 --enable-prefix-caching --trust-remote-code --language-model-only

Prefill

prompt tok/s by context length

8K context11.8K694 ms TTFT

16K context11K1,479 ms TTFT

32K context9,9303,252 ms TTFT

64K context8,2647,795 ms TTFT

128K context6,19320,780 ms TTFT

Sustained decode

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

Context	1	2	4	8	16	32	64	128
0	119.535 ms TTFT	229.745 ms TTFT	441.265 ms TTFT	805.498 ms TTFT	1,406.3170 ms TTFT	2,186.7247 ms TTFT	Over capacity	Over capacity
16K	111.389 ms TTFT	204.6129 ms TTFT	385.2184 ms TTFT	663.3330 ms TTFT	1,013.5601 ms TTFT	Over capacity	Over capacity	Over capacity
32K	104.5120 ms TTFT	184.0140 ms TTFT	340.7205 ms TTFT	549.9345 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity
64K	92.9205 ms TTFT	153.1256 ms TTFT	274.4374 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
128K	76.2542 ms TTFT	114.5597 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
248K	56.8907 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	11.8K	—	694 ms	—	—	1	—	—
Prefill	16K	1	Valid	11K	—	1,479 ms	—	—	1	—	—
Prefill	32K	1	Valid	9,930.0	—	3,252 ms	—	—	1	—	—
Prefill	64K	1	Valid	8,264.0	—	7,795 ms	—	—	1	—	—
Prefill	128K	1	Valid	6,193.0	—	20,780 ms	—	—	1	—	—
Sustained decode	0	1	Valid	119.5	119.7	35 ms	8.4 ms	1.0	0	3,584	30.0 s
Sustained decode	0	2	Valid	229.7	115.2	45 ms	8.7 ms	2.0	0	6,891	30.0 s
Sustained decode	0	4	Valid	441.2	110.9	65 ms	9.0 ms	4.0	0	13,232	30.0 s
Sustained decode	0	8	Valid	805.4	101.6	98 ms	9.8 ms	8.0	0	24,160	30.0 s
Sustained decode	0	16	Valid	1,406.3	89.2	170 ms	11.2 ms	16.0	0	42,192	30.0 s
Sustained decode	0	32	Valid	2,186.7	69.5	247 ms	14.4 ms	32.0	0	65,632	30.0 s
Sustained decode	0	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	0	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	1	Valid	111.3	111.4	89 ms	9.0 ms	1.0	0	3,338	30.0 s
Sustained decode	16K	2	Valid	204.6	102.4	129 ms	9.8 ms	2.0	0	6,136	30.0 s
Sustained decode	16K	4	Valid	385.2	96.4	184 ms	10.4 ms	4.0	0	11,556	30.0 s
Sustained decode	16K	8	Valid	663.3	82.7	330 ms	12.1 ms	8.0	0	19,896	30.0 s
Sustained decode	16K	16	Valid	1,013.5	62.8	601 ms	15.9 ms	16.0	0	30,400	30.0 s
Sustained decode	16K	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	32K	1	Valid	104.5	104.6	120 ms	9.6 ms	1.0	0	3,133	30.0 s
Sustained decode	32K	2	Valid	184.0	92.1	140 ms	10.9 ms	2.0	0	5,518	30.0 s
Sustained decode	32K	4	Valid	340.7	85.3	205 ms	11.7 ms	4.0	0	10,220	30.0 s
Sustained decode	32K	8	Valid	549.9	68.7	345 ms	14.6 ms	8.0	0	16,496	30.0 s
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	92.9	93.1	205 ms	10.7 ms	1.0	0	2,788	30.0 s
Sustained decode	64K	2	Valid	153.1	76.6	256 ms	13.0 ms	2.0	0	4,594	30.0 s
Sustained decode	64K	4	Valid	274.4	68.6	374 ms	14.6 ms	4.0	0	8,232	30.0 s
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	76.2	76.4	542 ms	13.1 ms	1.0	0	2,285	30.0 s
Sustained decode	128K	2	Valid	114.5	57.3	597 ms	17.5 ms	2.0	0	3,434	30.0 s
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	248K	1	Valid	56.8	57.1	907 ms	17.5 ms	1.0	0	1,705	30.0 s
Sustained decode	248K	2	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	248K	4	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	248K	8	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	248K	16	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	248K	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	248K	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	248K	128	Over capacity	—	—	—	—	—	—	—	—

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