Menu

Pareto Frontier: Throughput vs. Interactivity

Default view Showing the 4 most recently published submissions that match your filters.

Remember to review the Messaging Guidelines when using this material.

Chart Data Division: Standardized Model: DeepSeek-R1 # of Systems: 4 of 4 displayed Normalization: Off / Performance Only

100%
0.0100k200k300k400k500kSystem Throughput (Tokens/s) · higher is better100200300Interactivity (Tokens/s per User) · constraint: minimum or equalMLPerf Endpoints v1.0, DeepSeek-R1,Standardized division: GoogleGB300-NVL72_GB300-288GB_aarch64x72_TRT(Available, published June 26, 2026,provisioned power not reported); CoreWeaveNVIDIA GB200 NVL72 (Available, published June25, 2026, provisioned power not reported);NVIDIA GB200 NVL72 (Available, published June25, 2026, provisioned power not reported);NVIDIA GB300 NVL72 (Available, published June25, 2026, provisioned power not reported).Retrieved from endpoints.mlcommons.org onOctober 13, 2026. Results verified byMLCommons Association. The MLPerf name andlogo are registered and unregisteredtrademarks of MLCommons Association in theUnited States and other countries. All rightsreserved. Unauthorized use strictlyprohibited. See www.mlcommons.org for moreinformation.