Jesse Ellis
June 4, 2025 15:03
COREWEAVE has shown significant advances in AI performance by setting up a new MLPERF benchmark record using NVIDIA and IBM in partnership with NVIDIA GB200 Grace Blackwell Superchips.
COREWEAVE has set a breakthrough for MLPERF Training V5.0 benchmarks using NVIDIA and IBM in collaboration with NVIDIA GB200 Grace Blackwell Superchips. According to PR NEWSWIRE, this achievement uses almost 2,500 NVIDIA GB200 GPUs to indicate the largest submissions recorded on the benchmark.
Breakthrough performance
The submissions utilized the power of the 2,496 NVIDIA Blackwell GPU on the AI-OP Optimization Cloud Platform of COREWEAVE. Through this setting, COREWEAVE was able to complete the work within 27.3 minutes by achieving a breakthrough development in the benchmarking, especially the complex LLAMA 3.1 405B model. This performance is more than twice as much as similar submissions and emphasizes the function of the state -of -the -art infrastructure of the GB200 NVL72 architecture and COREWEAVE.
Implications for AI development
The effect of these results is important for the development of AI. It is important for the development of AI because the model development cycle is fast and the total cost of the AI project is reduced. COREWEAVE offers a competitive advantage in distributing and developing AI models to customers by reducing education time in half and optimizing cost efficiency.
COREWEAVE’s industry leadership
Peter Salanki, the chief technical officer and co -founder of COREWEAVE, emphasized the company’s promise to provide a high -performance cloud platform adjusted for the AI workload. The development of COREWEAVE is not theoretical. MLPERF Inference V5.0 and Training V5.0 Benchmark actually applies as proven by the main location.
The perception of COREWEAVE expands beyond the technical benchmark. It has been nominated among the most influential companies of Time100, and in 2024, forbes Cloud 100, and the influence of the technology industry is increasing.
About COREWEAVE
Founded in 2017, COREWEAVE operates a data center network throughout the United States and Europe, which specializes in AI-OP Optimization Cloud Solutions. The company is known as AI HyperScaler ™ and continues to push bags from accelerated computing for enterprises and AI Labs worldwide.
Image Source: Shutter Stock