入门指南排行榜Decode calculator模型OrganizationsReportsMatrix runs硬件基准评测市场租用ProAPI文档
语言
Actual Computer — Every computer, one endpoint

Speed test matrix

Laguna-XS-2.1-INT4 on H200 NVL 141GB

sglang · 0.5.6.post3.dev9269+g834400705 · INT4+INT86 contexts (8K–251K) × 8 concurrency levels (1–128) · prefill at 8K, 16K, 32K, 64K, 128K, 251K, 256K
by @keennaylmx dev importMeasured

Hardware

Type
Discrete GPU
GPU
H200 NVL
VRAM
141 GB
OS
linux

Engine

Engine
sglang
Version
0.5.6.post3.dev9269+g834400705
Quantization
INT4+INT8
Backend
cuda
Build
cuda-13.2
Commit
834400705f2d

Run

Served model
poolside
Revision
main
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_h200-poolside-sglang/bin/sglang" serve --model-path poolside/Laguna-XS-2.1-INT4 --served-model-name poolside --trust-remote-code --tp 1 --reasoning-parser poolside_v1 --tool-call-parser poolside_v1 --context-length 262144 --mem-fraction-static 0.851767 --enable-metrics --host 0.0.0.0 --port 8000 --api-key <redacted>

Prefill

prompt tok/s by context length
8K context33.2K247 ms TTFT
16K context32.3K502 ms TTFT
32K context30K1,074 ms TTFT
64K context25.5K2,521 ms TTFT
128K context18.5K6,924 ms TTFT
251K context12.1K20,784 ms TTFT
256K context12K21,389 ms TTFT

Sustained decode

total tok/s · rows are context length, columns are simultaneous requests
Context1248163264128
8K187.965 ms TTFT232.483 ms TTFT619.5104 ms TTFT970.7131 ms TTFT1,451.2196 ms TTFT2,633.3323 ms TTFTOver capacityOver capacity
16K184.888 ms TTFT251.9104 ms TTFT499.5131 ms TTFT838.7196 ms TTFT1,606.6329 ms TTFT2,383.0634 ms TTFTOver capacityOver capacity
32K179.3115 ms TTFT187.7138 ms TTFT556.2217 ms TTFT717.8358 ms TTFT1,317.5680 ms TTFTOver capacityOver capacityOver capacity
64K169.1168 ms TTFT213.4306 ms TTFT506.3350 ms TTFT564.6759 ms TTFTOver capacityOver capacityOver capacityOver capacity
128K152.6262 ms TTFT180.1436 ms TTFT377.2818 ms TTFTOver capacityOver capacityOver capacityOver capacityOver capacity
251K127.6501 ms TTFT197.0868 ms TTFTOver capacityOver capacityOver capacityOver capacityOver capacityOver 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.
PhaseContextRequests at onceStatusTotal tok/sPer request tok/sTTFTTime between tokensAverage active requestsCompleted requestsOutput tokensMeasured for
Prefill8K1Valid33.2K—247 ms——1——
Prefill16K1Valid32.3K—502 ms——1——
Prefill32K1Valid30K—1,074 ms——1——
Prefill64K1Valid25.5K—2,521 ms——1——
Prefill128K1Valid18.5K—6,924 ms——1——
Prefill251K1Valid12.1K—20,784 ms——1——
Prefill256K1Valid12K—21,389 ms——1——
Sustained decode8K1Valid187.9188.465 ms5.3 ms1.015,63530.0 s
Sustained decode8K2Valid232.4115.583 ms8.7 ms2.906,97130.0 s
Sustained decode8K4Valid619.5155.9104 ms6.4 ms4.0418,58430.0 s
Sustained decode8K8Valid970.7128.5131 ms7.8 ms10.3829,12130.0 s
Sustained decode8K16Valid1,451.291.4196 ms10.9 ms24.0043,53630.0 s
Sustained decode8K32Valid2,633.380.3323 ms12.5 ms36.0079,00830.0 s
Sustained decode8K64Over capacity————————
Sustained decode8K128Over capacity————————
Sustained decode16K1Valid184.8185.588 ms5.4 ms1.015,54330.0 s
Sustained decode16K2Valid251.9118.7104 ms8.4 ms14.827,55430.0 s
Sustained decode16K4Valid499.5137.2131 ms7.4 ms5.6414,98430.0 s
Sustained decode16K8Valid838.7105.2196 ms9.5 ms11.9025,16030.0 s
Sustained decode16K16Valid1,606.699.9329 ms10.0 ms16.0048,19230.0 s
Sustained decode16K32Valid2,383.073.3634 ms13.6 ms33.6071,48930.0 s
Sustained decode16K64Over capacity————————
Sustained decode16K128Over capacity————————
Sustained decode32K1Valid179.3180.2115 ms5.5 ms1.015,37830.0 s
Sustained decode32K2Valid187.790.5138 ms11.1 ms23.305,63130.0 s
Sustained decode32K4Valid556.2138.8217 ms7.2 ms4.1416,68730.0 s
Sustained decode32K8Valid717.890.2358 ms11.1 ms12.0021,52830.0 s
Sustained decode32K16Valid1,317.580.8680 ms12.4 ms17.9039,52030.0 s
Sustained decode32K32Over capacity————————
Sustained decode32K64Over capacity————————
Sustained decode32K128Over capacity————————
Sustained decode64K1Valid169.1170.3168 ms5.9 ms1.015,07430.0 s
Sustained decode64K2Valid213.4102.6306 ms9.8 ms9.806,40230.0 s
Sustained decode64K4Valid506.3126.2350 ms7.9 ms3.9415,18630.0 s
Sustained decode64K8Valid564.671.2759 ms14.0 ms12.0016,93630.0 s
Sustained decode64K16Over capacity————————
Sustained decode64K32Over capacity————————
Sustained decode64K64Over capacity————————
Sustained decode64K128Over capacity————————
Sustained decode128K1Valid152.6154.2262 ms6.5 ms1.014,57930.0 s
Sustained decode128K2Valid180.188.0436 ms11.4 ms8.605,40230.0 s
Sustained decode128K4Valid377.291.3818 ms11.0 ms4.5011,31630.0 s
Sustained decode128K8Over capacity————————
Sustained decode128K16Over capacity————————
Sustained decode128K32Over capacity————————
Sustained decode128K64Over capacity————————
Sustained decode128K128Over capacity————————
Sustained decode251K1Valid127.6130.1501 ms7.7 ms1.013,82630.0 s
Sustained decode251K2Valid197.095.5868 ms10.5 ms2.805,90930.0 s
Sustained decode251K4Over capacity————————
Sustained decode251K8Over capacity————————
Sustained decode251K16Over capacity————————
Sustained decode251K32Over capacity————————
Sustained decode251K64Over capacity————————
Sustained decode251K128Over capacity————————

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