Speed test matrix

Agents-A1-4B on RTX 3090 24GB

vllm · 0.32.0 · BF166 contexts (0–248K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 64K, 128K

by @Sandbergllm-inference-bench v0.8.0Measured Oct 31, 2026, 04:07 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: BF16
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 InternScience/Agents-A1-4B --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 context11.9K705 ms TTFT

16K context11K1,493 ms TTFT

32K context9,8043,318 ms TTFT

64K context8,3167,775 ms TTFT

128K context6,20520,780 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	86.544 ms TTFT	169.881 ms TTFT	329.5123 ms TTFT	616.3195 ms TTFT	1,113.0356 ms TTFT	1,811.6492 ms TTFT	Over capacity	Over capacity
16K	82.185 ms TTFT	155.493 ms TTFT	297.3135 ms TTFT	527.5225 ms TTFT	846.9417 ms TTFT	Over capacity	Over capacity	Over capacity
32K	78.2156 ms TTFT	143.2182 ms TTFT	269.4259 ms TTFT	453.8419 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity
64K	71.6208 ms TTFT	123.8275 ms TTFT	226.2449 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
128K	61.2396 ms TTFT	97.3520 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
248K	48.0831 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	—	705 ms	—	—	1	—	—
Prefill	16K	1	Valid	11K	—	1,493 ms	—	—	1	—	—
Prefill	32K	1	Valid	9,804.0	—	3,318 ms	—	—	1	—	—
Prefill	64K	1	Valid	8,316.0	—	7,775 ms	—	—	1	—	—
Prefill	128K	1	Valid	6,205.0	—	20,780 ms	—	—	1	—	—
Sustained decode	0	1	Valid	86.5	86.6	44 ms	11.5 ms	1.0	0	2,595	30.0 s
Sustained decode	0	2	Valid	169.8	85.0	81 ms	11.8 ms	2.0	0	5,092	30.0 s
Sustained decode	0	4	Valid	329.5	82.6	123 ms	12.1 ms	4.0	0	9,884	30.0 s
Sustained decode	0	8	Valid	616.3	77.4	195 ms	12.9 ms	8.0	0	18,491	30.0 s
Sustained decode	0	16	Valid	1,113.0	70.0	356 ms	14.3 ms	16.0	0	33,391	30.0 s
Sustained decode	0	32	Valid	1,811.6	56.7	492 ms	17.6 ms	32.0	0	54,336	30.0 s
Sustained decode	0	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	0	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	1	Valid	82.1	82.1	85 ms	12.2 ms	1.0	0	2,462	30.0 s
Sustained decode	16K	2	Valid	155.4	77.8	93 ms	12.9 ms	2.0	0	4,663	30.0 s
Sustained decode	16K	4	Valid	297.3	74.4	135 ms	13.4 ms	4.0	0	8,916	30.0 s
Sustained decode	16K	8	Valid	527.5	66.0	225 ms	15.2 ms	8.0	0	15,824	30.0 s
Sustained decode	16K	16	Valid	846.9	52.8	417 ms	18.9 ms	16.0	0	25,408	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	78.2	78.3	156 ms	12.8 ms	1.0	0	2,346	30.0 s
Sustained decode	32K	2	Valid	143.2	71.6	182 ms	14.0 ms	2.0	0	4,294	30.0 s
Sustained decode	32K	4	Valid	269.4	67.3	259 ms	14.9 ms	4.0	0	8,079	30.0 s
Sustained decode	32K	8	Valid	453.8	56.5	419 ms	17.7 ms	8.0	0	13,608	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	71.6	71.6	208 ms	14.0 ms	1.0	0	2,147	30.0 s
Sustained decode	64K	2	Valid	123.8	61.9	275 ms	16.1 ms	2.0	0	3,714	30.0 s
Sustained decode	64K	4	Valid	226.2	56.5	449 ms	17.7 ms	4.0	0	6,786	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	61.2	61.3	396 ms	16.3 ms	1.0	0	1,836	30.0 s
Sustained decode	128K	2	Valid	97.3	48.7	520 ms	20.6 ms	2.0	0	2,916	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	48.0	48.2	831 ms	20.7 ms	1.0	0	1,440	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.