Menu
Back previous/next report: k j

CoreWeave NVIDIA GB200 NVL72

Submitted by CoreWeaveon 2026-06-25

Result statement

CoreWeave ran DeepSeek-R1 on NVIDIA GB200 NVL72 and achieved a verified 441,740.3 tokens/second System TPS at 404.0ms TTFT P90 in the Standardized division of the Available category .

[1] MLPerf Endpoints v1.0 DeepSeek-R1; retrieved from endpoints.mlcommons.org on October 16, 2026. Result —, run 10: concurrency = 16,384 queries, TTFT P90 = 404.0ms, System TPS = 441,740.3 tokens/sec and interactivity = 27.0 tokens/sec per user. Result verified by MLCommons Association. The MLPerf name and logo are registered and unregistered trademarks of MLCommons Association in the United States and other countries. All rights reserved. Unauthorized use strictly prohibited. See www.mlcommons.org for more information.

Submission Info
CoreWeave
6535925d-129c-42f5-b5a8-5fb6211e0e4c
Submission Status PUBLISHED
2026-06-25
—
Division Standardized
Deployment Client over Network
Model
Model Name DeepSeek-R1
Division Standardized
Model Precision nvfp4
Model Link —
Calibration Link —
Model Notes —
Datasets
Dataset Name dsr1-perf
Dataset Type performance
Average Input Tokens —
Dataset Link —

System TPS vs. Tokens/s per User

0100k200k300k400kSystem TPS50100150200250Tokens/s per User

System TPS vs. Concurrency

0100k200k300k400kSystem TPS05,00010,00015,000# of queries

Time to First Token P90 vs. Concurrency

02.0k4.0k6.0k8.0k10kTTFT P90 (ms)05,00010,00015,000# of queries

Tokens/s per User vs. Concurrency

50100150200250Tokens/s per User05,00010,00015,000# of queries
SUT Summary
NVIDIA
NVIDIA GB200 NVL72
Available
System Category Datacenter · —
68x NVIDIA GB200
1
17
Max Supported Concurrency 16,384
Node Type ID 1
Node Count 17
Software
Ubuntu
Framework —
—
CUDA 13.2
595.71.05
Link to Container —
Operating System linux Ubuntu 24.04
Filesystem Not available
Other Software —
Software Notes —
Hardware

Processor

Neoverse-V2
1
144
VCPUs Per Processor 144

Accelerator

NVIDIA GB200
4
HBM3e
186GiB
NVLink 5 + NVSwitch (NVL72)
NVLink-C2C (Grace-Blackwell)

Host / Storage

956 GB
Memory Configuration —
Storage Capacity 14306 GB
Storage Type —

Networking

1x
NVLink + InfiniBand
Network Topology —

Other

Other Hardware —
Cooling —
Hardware Notes —
Run Data
Field NameRun 1Run 2Run 3Run 4Run 5Run 6Run 7Run 8Run 9Run 10
Run Date06/26/202606/25/202606/25/202606/25/202606/26/202606/25/202606/26/202606/25/202606/25/202606/25/2026
Concurrency1632482565121,0242,0484,0968,19216,384
System Tokens/Second4,017.47,057.310,485.746,676.278,791.9127,635.8193,265.1290,645.5389,090.6441,740.3
Tokens/Second per User251.1220.5218.5182.3153.9124.694.471.047.527.0
TTFT P90 (ms)10113.4262.4297.01671.11207.51290.41044.73073.95298.7404.0
Utilization0.9%1.6%2.4%10.6%17.8%28.9%43.8%65.8%88.1%100.0%
Configuration Summarydsr1-gb200x72-pareto-nvl72-c16 · submissiondsr1-gb200x72-pareto-nvl72-c32 · submissiondsr1-gb200x72-pareto-nvl72-c32 · submission · c48dsr1-gb200x72-pareto-nvl72-c256 · submissiondsr1-gb200x72-pareto-nvl72-c512 · submissiondsr1-gb200x72-pareto-nvl72-c1024 · submissiondsr1-gb200x72-pareto-nvl72-c2048 · submissiondsr1-gb200x72-pareto-nvl72-c4096 · submissiondsr1-gb200x72-pareto-nvl72-c8192 · submissiondsr1-gb200x72-pareto-nvl72-c16384 · submission
Config Typesubmissionsubmissionsubmissionsubmissionsubmissionsubmissionsubmissionsubmissionsubmissionsubmission
Config Region——————————
Target QPS——————————
Result Statecompletecompletecompletecompletecompletecompletecompletecompletecompletecomplete
Measured Accuracy Score——————————
Benchmark Versiond0b1154d0b1154d0b1154d0b1154d0b1154d0b1154d0b1154d0b1154d0b1154d0b1154
Result Version0.1.00.1.00.1.00.1.00.1.00.1.00.1.00.1.00.1.00.1.0
Archive URIgs://mlperf-endpoints-blob/runs/543bf208-1752-456a-9023-2b9def7fc888/archive.tar.gzgs://mlperf-endpoints-blob/runs/ac882870-e1ae-4144-bbf1-0607920db278/archive.tar.gzgs://mlperf-endpoints-blob/runs/b5fc9d3e-d626-4d35-ba82-ca3b016187a9/archive.tar.gzgs://mlperf-endpoints-blob/runs/52394b12-bfe2-4413-90a3-5786bda92e14/archive.tar.gzgs://mlperf-endpoints-blob/runs/934f515f-0f69-4583-89de-9bea66f01af1/archive.tar.gzgs://mlperf-endpoints-blob/runs/46db041f-c562-4d54-9e02-64266005dcef/archive.tar.gzgs://mlperf-endpoints-blob/runs/d9b32918-a093-4434-9c50-44f1dc5e41e0/archive.tar.gzgs://mlperf-endpoints-blob/runs/45ea80ed-771f-4da4-98c3-cd4259354122/archive.tar.gzgs://mlperf-endpoints-blob/runs/76e09625-871d-43d7-9a92-a50f4f523d96/archive.tar.gzgs://mlperf-endpoints-blob/runs/6646e4e5-afd0-4442-a52c-fd2f2aa06315/archive.tar.gz
QPS1.10901.87502.783312.367820.954533.839151.513877.4180103.9316117.0239
Total Output Tokens15,576,84428,605,48928,631,907185,304,678311,716,324499,015,460756,721,1911,144,291,8531,413,627,5131,710,733,049
Average Output Tokens3,622.523,763.883,767.363,774.033,760.153,771.853,751.723,754.243,743.723,774.79
Run Duration (s)3,877.394,053.302,730.583,970.003,956.203,909.683,915.463,937.073,633.163,872.71
Total Queries4,3007,6007,60049,10082,900132,300201,700304,800377,600453,200
Total Queries Issued4,3007,6007,60049,10082,900132,300201,700304,800377,600453,200
Total Queries Failed0000000000
Time to First Token (TTFT) (ms)
Minimum139.7146.1150.9157.1170.7168.3200.2200.5199.5176.3
Average3175.9199.6225.9680.9616.1584.0621.61371.82997.11464.6
P50237.5171.8187.3336.4422.5317.2367.7371.7514.2329.0
P9010113.4262.4297.01671.11207.51290.41044.73073.95298.7404.0
P9514811.8290.9316.52262.01679.01680.21362.44347.914163.15770.4
P9924302.7336.0364.03427.93113.82900.36120.617842.845355.533413.1
P99935571.52438.42807.04534.34983.88406.915793.031734.163584.243135.4
Maximum46355.12863.03151.85743.75004.99035.217665.434197.367197.344868.4
Time Per Output Token (TPOT) (ms)
Minimum0.000.840.012.622.371.971.962.335.286.53
Average3.184.524.545.356.367.8810.413.619.434.8
P503.144.474.495.306.307.8110.413.619.635.6
P903.594.985.015.927.048.7511.615.220.638.2
P953.675.095.106.047.188.9211.915.520.838.8
P993.825.255.276.267.439.2312.316.121.239.7
P9994.005.485.506.507.739.6112.816.821.640.6
Maximum4.325.595.647.048.2710.514.218.022.943.4
Query Latency (ms)
Minimum793.51069.91001.21122.11196.41371.91494.01915.53176.84847.7
Average14381.216772.116880.420351.123941.929547.838755.351260.074849.5131201.6
P5010005.69561.19727.211929.713933.117172.422353.829909.644498.374738.7
P9032199.341480.441297.449578.258127.472564.096045.5125891.4182685.4324380.7
P9545049.158435.758475.170747.782611.8102971.4135723.4178905.9257187.4457239.2
P9961756.583988.382490.699282.5117924.4145864.1193059.2253234.2376737.6671764.0
P99972460.791330.790509.5107746.4128241.0158353.6209822.7276608.7392739.4727381.0
Maximum92203.297798.193067.2120302.8154702.5182600.6261548.6360922.4408893.9754525.7