Hardik Shah
2026.06.16 15:05

๐Ÿ“ข ๐‰๐”๐’๐“ ๐ˆ๐: $Coreweave(CRWV.US) CoreWeave Trains DeepSeek-V3 671B in Just 2.02 Minutes in MLPerf v6.0 - $NVIDIA(NVDA.US)

๐Ÿ‘‰ ๐Š๐ž๐ฒ ๐‡๐ข๐ ๐ก๐ฅ๐ข๐ ๐ก๐ญ๐ฌ:

โžค ๐‚๐จ๐ซ๐ž๐–๐ž๐š๐ฏ๐ž trained ๐ƒ๐ž๐ž๐ฉ๐’๐ž๐ž๐ค-๐•๐Ÿ‘ ๐Ÿ”๐Ÿ•๐Ÿ๐ in just ๐Ÿ.๐ŸŽ๐Ÿ minutes.

โžค Record achieved using ๐Ÿ–,๐Ÿ๐Ÿ—๐Ÿ ๐๐•๐ˆ๐ƒ๐ˆ๐€ ๐†๐๐Ÿ‘๐ŸŽ๐ŸŽ ๐๐•๐‹๐Ÿ•๐Ÿ GPUs across ๐Ÿ,๐ŸŽ๐Ÿ’๐Ÿ– nodes.

โžค CoreWeave posted the ๐Ÿ๐š๐ฌ๐ญ๐ž๐ฌ๐ญ Closed/Available-cloud results in MLPerf Training v6.0.

โžค DeepSeek-V3 training completed in ๐Ÿ‘.๐ŸŽ๐Ÿ— minutes on ๐Ÿ’,๐ŸŽ๐Ÿ—๐Ÿ” GPUs.

โžค DeepSeek-V3 training completed in ๐Ÿ“.๐Ÿ“๐Ÿ’ minutes on ๐Ÿ,๐ŸŽ๐Ÿ’๐Ÿ– GPUs.

โžค CoreWeave was the ๐จ๐ง๐ฅ๐ฒ ๐ฌ๐ฎ๐›๐ฆ๐ข๐ญ๐ญ๐ž๐ซ scaling GB300 beyond ๐Ÿ,๐ŸŽ๐Ÿ’๐Ÿ– GPUs.

โžค Llama-3.1-405B reached benchmark quality in ๐Ÿ—.๐Ÿ•๐Ÿ• minutes on ๐Ÿ’,๐ŸŽ๐Ÿ—๐Ÿ” GPUs.

โžค GPT-OSS-20B trained in ๐Ÿ๐Ÿ”.๐Ÿ—๐Ÿ– minutes on a ๐Ÿ”๐Ÿ’-GPU HGX B200 cluster.

โžค Llama-3.1-8B trained in ๐Ÿ๐Ÿ”.๐Ÿ“๐Ÿ’ minutes on the same ๐Ÿ”๐Ÿ’-GPU cluster.

โžค CoreWeave said all results used its ๐ฉ๐ซ๐จ๐๐ฎ๐œ๐ญ๐ข๐จ๐ง ๐œ๐ฅ๐จ๐ฎ๐ infrastructure available to customers

The copyright of this article belongs to the original author/organization.

The views expressed herein are solely those of the author and do not reflect the stance of the platform. The content is intended for investment reference purposes only and shall not be considered as investment advice. Please contact us if you have any questions or suggestions regarding the content services provided by the platform.

ยฉ 2026 Longbridge๏ฝœDisclaimer