Speed test matrix

Qwen3.8-27B-INT4 on RTX 3090 24GB

vllm · 0.32.0 · INT4 W4A16 g1283 contexts (0–24K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 24K

by @Sandbergllm-inference-bench v0.8.0Measured Oct 28, 2026, 23:55 UTCEdited Oct 29, 2026, 12:41 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: INT4 W4A16 g128
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/Qwen3.8-27B-INT4 --max-model-len 32768 --gpu-memory-utilization 0.94 --max-num-seqs 54 --enable-prefix-caching --trust-remote-code --language-model-only --max-cudagraph-capture-size 16

Prefill

prompt tok/s by context length

8K context2,1793,779 ms TTFT

16K context2,0887,790 ms TTFT

24K context2,01212,075 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	48.281 ms TTFT	89.2414 ms TTFT	171.1266 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
16K	45.3400 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
24K	44.1515 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	2,179.0	—	3,779 ms	—	—	1	—	—
Prefill	16K	1	Valid	2,088.0	—	7,790 ms	—	—	1	—	—
Prefill	24K	1	Valid	2,012.0	—	12,075 ms	—	—	1	—	—
Sustained decode	0	1	Valid	48.2	48.2	81 ms	20.8 ms	1.0	0	1,444	30.0 s
Sustained decode	0	2	Valid	89.2	44.6	414 ms	22.4 ms	2.0	0	2,674	30.0 s
Sustained decode	0	4	Valid	171.1	42.9	266 ms	23.3 ms	4.0	0	5,132	30.0 s
Sustained decode	0	8	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	0	16	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	0	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	0	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	0	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	1	Valid	45.3	45.3	400 ms	22.1 ms	1.0	0	1,358	30.0 s
Sustained decode	16K	2	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	4	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	8	Over capacity	—	—	—	—	—	—	—	—
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	24K	1	Valid	44.1	44.1	515 ms	22.7 ms	1.0	0	1,322	30.0 s
Sustained decode	24K	2	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	24K	4	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	24K	8	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	24K	16	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	24K	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	24K	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	24K	128	Over capacity	—	—	—	—	—	—	—	—

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