Apr, 2025 : CoreWeave Claims AI Inference Record with NVIDIA GB200


CoreWeave Claims AI Inference Record with NVIDIA GB200

📅 - U.S.-based AI hyperscaler CoreWeaveclaims to have set a new benchmark for AI inference performance using NVIDIAs latest GB200 Grace Blackwell Superchips, according to results released through the MLPerf v5.0 benchmark suite. The company reports achieving 800 tokens per second (TPS) on the open-source Llama 3.1 405B model using a single instance equipped with two NVIDIA Grace CPUs and four NVIDIA Blackwell GPUs.

The MLPerf benchmarks are widely recognized as an industry standard for evaluating machine learning system performance under real-world conditions, particularly focusing on inference speed - how quickly a trained model can produce usable outputs.

Inference speed is a key factor in [...][... Check source for end of article ...]
Tags: AI

Reads: 117 | Category: General | Source: Hosting Jurnalist : Hosting Jurnalist | Author:
URL source: https://hostingjournalist.com/news/coreweave-claims-ai-inference-record-with-nvidia-gb200
Want to add a website news or press release ? Just do it, it's free! Use add web hosting news!