Speed test matrix

gemma-4-12B-it on RTX PRO 6000 Blackwell Server Edition 96GB

sglang · 0.5.22.dev236+g1093c501d · BF166 contexts (8K–251K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 64K, 128K, 256K

by @keennaylmx dev importMeasured Oct 3, 2026, 21:08 UTC

Hardware

Type: Discrete GPU
GPU: RTX PRO 6000 Blackwell Server Edition
VRAM: 96 GB
OS: linux

Engine

Engine: sglang
Version: 0.5.22.dev236+g1093c501d
Quantization: BF16
Backend: cuda
Repository: github.com/sgl-project/sglang
Build: cuda-13.2
Commit: 1093c501dfef

Run

Served model: gemma
Revision: main
Benchmark tool: llm-inference-bench v0.7.6
Decode test: Fixed duration
Prefill mode: Measured during decode warm-up
Time per measurement: 30 s
Max output: 4,096 tokens

Launch command

"$HOME/env_rtxpro6k-google-sglang-1093c501dfef/bin/sglang" serve --model-path google/gemma-4-12B-it --served-model-name gemma --trust-remote-code --tp 1 --reasoning-parser gemma4 --tool-call-parser gemma4 --context-length 262144 --mem-fraction-static 0.78 --enable-metrics --host 0.0.0.0 --port 8000 --api-key <redacted> --enable-unified-memory --mm-process-config '{"image":{"max_soft_tokens":1120},"video":{"max_soft_tokens":1120}}' --attention-backend triton --bf16-gemm-backend torch

Prefill

prompt tok/s by context length

8K context9,728843 ms TTFT

16K context8,5471,900 ms TTFT

32K context6,9224,672 ms TTFT

64K context4,81013,413 ms TTFT

128K context2,59649,646 ms TTFT

256K context1,372187,750 ms TTFT

Sustained decode

total tok/s · rows are context length, columns are simultaneous requests

Context	1	2	4	8	16	32	64	128
8K	50.0818 ms TTFT	100.1155 ms TTFT	Looped	Looped	Over capacity	Over capacity	Over capacity	Over capacity
16K	Looped	Looped	186.7132 ms TTFT	Looped	Over capacity	Over capacity	Over capacity	Over capacity
32K	42.097 ms TTFT	84.6187 ms TTFT	Looped	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
64K	34.7155 ms TTFT	70.2302 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
128K	25.6270 ms TTFT	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity
251K	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity	Over capacity

Looped: the model kept repeating the same text, so the benchmark discarded this measurement. 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	9,728.0	—	843 ms	—	—	1	—	—
Prefill	16K	1	Valid	8,547.0	—	1,900 ms	—	—	1	—	—
Prefill	32K	1	Valid	6,922.0	—	4,672 ms	—	—	1	—	—
Prefill	64K	1	Valid	4,810.0	—	13,413 ms	—	—	1	—	—
Prefill	128K	1	Valid	2,596.0	—	49,646 ms	—	—	1	—	—
Prefill	256K	1	Valid	1,372.0	—	187,750 ms	—	—	1	—	—
Sustained decode	8K	1	Valid	50.0	50.0	818 ms	20.0 ms	1.0	0	1,500	30.0 s
Sustained decode	8K	2	Valid	100.1	50.0	155 ms	20.0 ms	2.0	0	3,002	30.0 s
Sustained decode	8K	4	Looped	—	—	—	—	—	0	0	—
Sustained decode	8K	8	Looped	—	—	—	—	—	0	0	—
Sustained decode	8K	16	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	8K	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	8K	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	8K	128	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	16K	1	Looped	—	—	—	—	—	0	0	—
Sustained decode	16K	2	Looped	—	—	—	—	—	0	0	—
Sustained decode	16K	4	Valid	186.7	46.8	132 ms	21.4 ms	4.0	0	5,600	30.0 s
Sustained decode	16K	8	Looped	—	—	—	—	—	0	0	—
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	42.0	42.1	97 ms	23.8 ms	1.0	0	1,261	30.0 s
Sustained decode	32K	2	Valid	84.6	42.4	187 ms	23.6 ms	2.0	0	2,536	30.0 s
Sustained decode	32K	4	Looped	—	—	—	—	—	0	0	—
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	34.7	34.7	155 ms	28.8 ms	1.0	0	1,039	30.0 s
Sustained decode	64K	2	Valid	70.2	35.1	302 ms	28.5 ms	2.0	0	2,104	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	Valid	25.6	25.6	270 ms	39.1 ms	1.0	0	767	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	251K	1	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	251K	2	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	251K	4	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	251K	8	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	251K	16	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	251K	32	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	251K	64	Over capacity	—	—	—	—	—	—	—	—
Sustained decode	251K	128	Over capacity	—	—	—	—	—	—	—	—

Looped: the model kept repeating the same text, so the benchmark discarded this measurement. Over capacity: not run because this many requests at this context would exceed the server's KV cache.