Inf2 instances, optimized for generative AI models, are now generally available globally across 8 AWS Regions, providing high performance at the lowest cost for tasks like text summarization, code generation, and image/video processing.

<div>
<p>The article announces the global availability of Amazon EC2 Inf2 instances, which are optimized for generative AI workloads such as text summarization, code generation, video/image generation, speech recognition, and personalization.</p>
<p>Specifically, the article covers:</p>
<ul>
<li>Inf2 instances deliver high performance at the lowest cost on Amazon EC2 for generative AI models</li>
<li>Support for scale-out distributed inference with NeuronLink interconnect</li>
<li>High performance with up to 2.3 petaflops and 384 GB accelerator memory</li>
<li>Up to 40% better price-performance than comparable EC2 instances</li>
<li>Native integration with popular machine learning frameworks via AWS Neuron SDK</li>
<li>Available in four sizes across 8 AWS Regions as On-Demand, Reserved, Spot Instances, or Savings Plans</li>
</ul>
</div>


Amazon EC2 Inf2 instances, optimized for generative AI, now available globally

Related articles