CoreWeave, the AI Hyperscaler, has once again raised the bar in AI inference performance. The company announced its latest MLPerf v5.0 results, setting a new industry benchmark with NVIDIA GB200 Grace Blackwell Superchips. This achievement solidifies CoreWeave’s position as a leader in high-performance AI infrastructure.

CoreWeave’s Record-Breaking Performance

Using a CoreWeave instance powered by NVIDIA GB200—which includes two NVIDIA Grace CPUs and four NVIDIA Blackwell GPUs—the company delivered an impressive 800 tokens per second (TPS) on the Llama 3.1 405B model. This model ranks among the largest open-source AI models available.

AI Authority Trend: CoreWeave Appoints Jean English as New Chief Marketing Officer to Spearhead AI Growth

Peter Salanki, Chief Technology Officer at CoreWeave, emphasized the company’s commitment to innovation. ” CoreWeave is committed to building high-performance cloud infrastructure tailored for large-model inference, ensuring speed, scalability, and efficiency. These benchmark MLPerf results reinforce our position as the preferred cloud provider for top AI labs and enterprises,” he stated.

Enhancing Performance with NVIDIA H200

Beyond its success with the NVIDIA GB200, CoreWeave also showcased groundbreaking results for NVIDIA H200 GPU instances. The company achieved 33,000 TPS on the Llama 2 70B model, marking a 40% increase in throughput compared to NVIDIA H100 instances. This significant improvement underscores CoreWeave’s ability to push the boundaries of AI inferencing.

Leading the AI Cloud Infrastructure Revolution

CoreWeave continues to drive innovation in AI cloud services. This year, it became the first to offer general availability of NVIDIA GB200 NVL72-based instances. Previously, the company was among the pioneers to introduce NVIDIA H100 and H200 GPUs, as well as one of the first to demonstrate NVIDIA GB200 NVL72 technology.

AI Authority Trend: CoreWeave and Bulk Infrastructure Partner for Major NVIDIA AI Deployment in Europe

MLPerf Inference serves as an industry-standard benchmark for measuring machine learning performance in real-world scenarios. Faster inference speeds translate directly to enhanced user experiences, reinforcing CoreWeave’s role in advancing AI-driven applications.

FAQs

1. What makes CoreWeave’s NVIDIA GB200-based cloud instances unique?

CoreWeave’s NVIDIA GB200 instances provide unmatched AI inferencing power, featuring two Grace CPUs and four Blackwell GPUs. These configurations ensure ultra-fast processing speeds and high efficiency for large AI models.

2. How does CoreWeave compare to other cloud providers in AI inference?

CoreWeave consistently outperforms competitors by optimizing its cloud infrastructure for AI workloads. Its MLPerf v5.0 results confirm its leadership in AI inferencing, with significant performance advantages over traditional cloud services.

3. Why are MLPerf benchmarks important for AI applications?

MLPerf benchmarks offer a standardized way to measure AI model performance. They help enterprises and researchers choose the best infrastructure for deploying AI applications efficiently and at scale.

AI Authority Trend: CoreWeave Announces Agreement with OpenAI to Deliver AI Infrastructure

To share your insights, please write to us at sudipto@intentamplify.com