Speed test matrix

Ministral-3-3B-Instruct-2512-FP8 on RTX 3060 12GB

vllm · 0.32.0 · FP86 contexts (0–208K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 64K

by @Sandbergllm-inference-bench v0.8.0Measured Oct 31, 2026, 07:20 UTC

Hardware

Type: Discrete GPU
GPU: RTX 3060
VRAM: 12 GB
CPU: Intel Xeon E5-2680 v4 (14C)
RAM: 32 GB
OS: Ubuntu 24.04.5

Engine

Engine: vllm
Version: 0.32.0
Quantization: FP8
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 3060 12GB, stock 170 W limit, PCIe Gen3 x16 | Intel Xeon E5-2680 v4 14C | 32 GB DDR4-2400 | Ubuntu 24.04.5, NVIDIA driver 580.178.04
vllm serve unsloth/Ministral-3-3B-Instruct-2512-FP8 --max-model-len 221184 --gpu-memory-utilization 0.94 --max-num-seqs 128 --enable-prefix-caching --trust-remote-code --language-model-only --kv-cache-dtype int4_per_token_head --attention-backend TRITON_ATTN

Prefill

prompt tok/s by context length

8K context1,8284,557 ms TTFT

16K context1,21113,624 ms TTFT

32K context729.045,068 ms TTFT

64K context408.0160,757 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	67.316,139 ms TTFT	129.315,192 ms TTFT	238.9135 ms TTFT	422.6164 ms TTFT	696.2202 ms TTFT	Over capacity	Over capacity	Over capacity
16K	42.9131 ms TTFT	68.9207 ms TTFT	99.8302 ms TTFT	124.0412 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity
32K	31.4189 ms TTFT	46.4272 ms TTFT	61.7360 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
64K	20.5Requests queued324 ms TTFT	27.9Requests queued412 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
128K	11.8Requests queued1,016 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
208K	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity

Requests queued: the server could not run all requests at once, so some waited in a queue. 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	1,828.0	—	4,557 ms	—	—	1	—	—
Prefill	16K	1	Valid	1,211.0	—	13,624 ms	—	—	1	—	—
Prefill	32K	1	Valid	729.0	—	45,068 ms	—	—	1	—	—
Prefill	64K	1	Valid	408.0	—	160,757 ms	—	—	1	—	—
Sustained decode	0	1	Valid	67.3	67.6	16,139 ms	14.8 ms	1.0	0	2,019	30.0 s
Sustained decode	0	2	Valid	129.3	65.1	15,192 ms	15.4 ms	2.0	0	3,876	30.0 s
Sustained decode	0	4	Valid	238.9	60.8	135 ms	16.4 ms	4.0	0	7,164	30.0 s
Sustained decode	0	8	Valid	422.6	54.4	164 ms	18.4 ms	8.0	0	12,672	30.0 s
Sustained decode	0	16	Valid	696.2	45.6	202 ms	21.9 ms	16.0	0	20,880	30.0 s
Sustained decode	0	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	0	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	0	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	1	Valid	42.9	43.0	131 ms	23.2 ms	1.0	0	1,288	30.0 s
Sustained decode	16K	2	Valid	68.9	34.5	207 ms	29.0 ms	2.0	0	2,064	30.0 s
Sustained decode	16K	4	Valid	99.8	25.0	302 ms	39.9 ms	4.0	0	2,992	30.0 s
Sustained decode	16K	8	Valid	124.0	15.5	412 ms	64.3 ms	8.0	0	3,720	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	31.4	31.4	189 ms	31.8 ms	1.0	0	942	30.0 s
Sustained decode	32K	2	Valid	46.4	23.2	272 ms	43.1 ms	2.0	0	1,390	30.0 s
Sustained decode	32K	4	Valid	61.7	15.5	360 ms	64.6 ms	4.0	0	1,848	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	Requests queued	20.5	20.5	324 ms	48.8 ms	1.0	0	614	30.0 s
Sustained decode	64K	2	Requests queued	27.9	13.9	412 ms	71.8 ms	2.0	0	836	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	Requests queued	11.8	11.9	1,016 ms	83.9 ms	1.0	0	352	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	208K	1	Over capacity	—	—	—	—	0.4	0	0	30.0 s
Sustained decode	208K	2	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	208K	4	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	208K	8	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	208K	16	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	208K	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	208K	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	208K	128	Over capacity	—	—	—	—	—	—	—	—

Requests queued: the server could not run all requests at once, so some waited in a queue. Over capacity: not run because this many requests at this context would exceed the server's KV cache.