Speed test matrix

Qwen3-14B-AWQ on RTX 3090 24GB

vllm · 0.32.0 · AWQ3 contexts (0–32K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K

by @Sandbergllm-inference-bench v0.8.0Measured Oct 29, 2026, 15:37 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: AWQ
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. KV cache: int4_per_token_head
# 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 Qwen/Qwen3-14B-AWQ --max-model-len 40960 --gpu-memory-utilization 0.94 --max-num-seqs 128 --enable-prefix-caching --trust-remote-code --kv-cache-dtype int4_per_token_head --attention-backend TRITON_ATTN

Prefill

prompt tok/s by context length

8K context2,3313,539 ms TTFT

16K context1,6489,918 ms TTFT

32K context1,05031,004 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	73.940 ms TTFT	143.970 ms TTFT	270.774 ms TTFT	488.879 ms TTFT	824.9117 ms TTFT	1,407.4170 ms TTFT	Over capacity	Over capacity
16K	52.5154 ms TTFT	82.1171 ms TTFT	129.2221 ms TTFT	183.0308 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity
32K	40.6232 ms TTFT	56.9280 ms TTFT	84.3367 ms TTFT	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	2,331.0	—	3,539 ms	—	—	1	—	—
Prefill	16K	1	Valid	1,648.0	—	9,918 ms	—	—	1	—	—
Prefill	32K	1	Valid	1,050.0	—	31,004 ms	—	—	1	—	—
Sustained decode	0	1	Valid	73.9	74.4	40 ms	13.4 ms	1.0	0	2,217	30.0 s
Sustained decode	0	2	Valid	143.9	72.5	70 ms	13.8 ms	2.0	0	4,318	30.0 s
Sustained decode	0	4	Valid	270.7	68.6	74 ms	14.6 ms	4.0	0	8,120	30.0 s
Sustained decode	0	8	Valid	488.8	62.5	79 ms	16.0 ms	8.0	0	14,664	30.0 s
Sustained decode	0	16	Valid	824.9	53.4	117 ms	18.7 ms	16.0	0	24,750	30.0 s
Sustained decode	0	32	Valid	1,407.4	46.3	170 ms	21.6 ms	32.0	0	42,208	30.0 s
Sustained decode	0	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	0	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	1	Valid	52.5	52.6	154 ms	19.0 ms	1.0	0	1,576	30.0 s
Sustained decode	16K	2	Valid	82.1	41.2	171 ms	24.3 ms	2.0	0	2,463	30.0 s
Sustained decode	16K	4	Valid	129.2	32.5	221 ms	30.8 ms	4.0	0	3,876	30.0 s
Sustained decode	16K	8	Valid	183.0	22.9	308 ms	43.6 ms	8.0	0	5,488	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	40.6	40.7	232 ms	24.6 ms	1.0	0	1,219	30.0 s
Sustained decode	32K	2	Valid	56.9	28.5	280 ms	35.1 ms	2.0	0	1,706	30.0 s
Sustained decode	32K	4	Valid	84.3	21.1	367 ms	47.4 ms	4.0	0	2,528	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	—	—	—	—	—	—	—	—

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