Speed test matrix

Qwen3.5-4B-AWQ-4bit on RTX 3060 12GB

vllm · 0.32.0 · AWQ 4-bit6 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, 18:12 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: AWQ 4-bit
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: MTP, 2 tokens lookahead. KV cache: fp8
# 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 cyankiwi/Qwen3.5-4B-AWQ-4bit --max-model-len 262144 --gpu-memory-utilization 0.94 --max-num-seqs 128 --enable-prefix-caching --trust-remote-code --language-model-only --kv-cache-dtype fp8 --speculative-config {"method":"mtp","num_speculative_tokens":2}

Prefill

prompt tok/s by context length

8K context2,5573,186 ms TTFT

16K context2,3456,874 ms TTFT

32K context2,02915,817 ms TTFT

64K context1,59540,162 ms TTFT

128K context1,124113,842 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	83.0111 ms TTFT	166.8133 ms TTFT	304.3258 ms TTFT	488.7385 ms TTFT	737.3663 ms TTFT	578.2Requests queued1,108 ms TTFT	Over capacity	Over capacity
16K	83.5684 ms TTFT	166.41,238 ms TTFT	281.12,069 ms TTFT	383.43,184 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity
32K	88.71,450 ms TTFT	156.92,477 ms TTFT	248.44,133 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
64K	76.81,970 ms TTFT	135.33,632 ms TTFT	206.36,137 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
128K	67.22,655 ms TTFT	111.64,771 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
248K	48.8Requests queued4,058 ms TTFT	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	2,557.0	—	3,186 ms	—	—	1	—	—
Prefill	16K	1	Valid	2,345.0	—	6,874 ms	—	—	1	—	—
Prefill	32K	1	Valid	2,029.0	—	15,817 ms	—	—	1	—	—
Prefill	64K	1	Valid	1,595.0	—	40,162 ms	—	—	1	—	—
Prefill	128K	1	Valid	1,124.0	—	113,842 ms	—	—	1	—	—
Sustained decode	0	1	Valid	83.0	84.1	111 ms	11.9 ms	1.0	0	2,490	30.0 s
Sustained decode	0	2	Valid	166.8	84.9	133 ms	11.8 ms	2.0	0	5,002	30.0 s
Sustained decode	0	4	Valid	304.3	76.8	258 ms	13.0 ms	4.0	0	9,125	30.0 s
Sustained decode	0	8	Valid	488.7	61.9	385 ms	16.1 ms	8.0	0	14,655	30.0 s
Sustained decode	0	16	Valid	737.3	46.8	663 ms	21.4 ms	16.0	0	22,117	30.0 s
Sustained decode	0	32	Requests queued	578.2	26.8	1,108 ms	37.4 ms	24.9	0	17,314	29.9 s
Sustained decode	0	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	0	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	1	Valid	83.5	84.3	684 ms	11.9 ms	1.0	0	2,504	30.0 s
Sustained decode	16K	2	Valid	166.4	84.0	1,238 ms	11.9 ms	2.0	0	4,990	30.0 s
Sustained decode	16K	4	Valid	281.1	70.5	2,069 ms	14.2 ms	4.0	0	8,426	30.0 s
Sustained decode	16K	8	Valid	383.4	47.0	3,184 ms	21.3 ms	8.0	0	11,501	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	88.7	89.8	1,450 ms	11.1 ms	1.0	0	2,660	30.0 s
Sustained decode	32K	2	Valid	156.9	78.8	2,477 ms	12.7 ms	2.0	0	4,706	30.0 s
Sustained decode	32K	4	Valid	248.4	61.1	4,133 ms	16.4 ms	4.0	0	7,449	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	Valid	76.8	78.3	1,970 ms	12.8 ms	1.0	0	2,303	30.0 s
Sustained decode	64K	2	Valid	135.3	69.1	3,632 ms	14.5 ms	2.0	0	4,059	30.0 s
Sustained decode	64K	4	Valid	206.3	51.6	6,137 ms	19.4 ms	4.0	0	6,181	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	67.2	68.1	2,655 ms	14.7 ms	1.0	0	2,015	30.0 s
Sustained decode	128K	2	Valid	111.6	57.2	4,771 ms	17.5 ms	2.0	0	3,348	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	Requests queued	48.8	55.3	4,058 ms	18.1 ms	0.9	0	1,463	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	—	—	—	—	—	—	—	—

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.