Marktechpost AI introduces GPU leaderboard for cloud inference pricing and companion chrome extension

Marktechpost has released GPU Leaderboard, a free tool designed to compare cloud GPU pricing for AI inference, alongside a related Chrome extension, GPU Leaderboard: Deploy It. The dashboard functions as a pricing comparison engine, pulling on-demand and spot rates directly from various providers' pricing pages. Its core purpose is to help users quickly determine the necessary GPU and the most cost-effective provider. The initial version allows users to specify an accelerator, GPU quantity, duration, and utilization, providing a precise cost estimate across all listed providers. For example, a calculation for 8xH100 running for 72 hours at 90% utilization resulted in a cheapest price of $967.68 at Latitude, with a median cost across 15 providers being $1,716.48, while CoreWeave showed the highest cost at $3,545.28. The tool also highlights the difference between the quoted rate and the actual cost, noting that idle time billing can increase the effective rate from $1.68 to $1.87 per GPU-hour. The accompanying Chrome extension integrates this pricing data directly into developer workflows, functioning within three locations, and can be accessed via gpuleaderboard.com.