Marktechpost has launched GPU Leaderboard, a free cloud GPU price comparison dashboard for AI inference that aggregates on-demand and spot pricing from 24 providers' official pages, ranking 201 live GPU offers by cost per GPU-hour. The platform features four views: an inference pricing calculator for full-run costs, a filterable leaderboard ranked by $/GPU-hr, a cheapest inference floor per GPU, and a provider directory with instance counts, regions, incident status, and price-check history. The calculator accepts accelerator type, GPU count, duration, and utilization to price exact workloads across providers; an optional model field narrows results. Using the default example of 8×H100 GPUs for 72 hours at 90% utilization (576 GPU-hours), the cheapest quote is $967.68 from Latitude at $1.68/GPU-hr, the median across 15 providers is $1,716.48, and the highest is $3,545.28 at CoreWeave. The tool also computes a "real" cost per used GPU-hour accounting for idle billing, raising the Latitude rate to $1.87 at 90% utilization. The companion Chrome extension, GPU Leaderboard: Deploy It, integrates this data into developer workflows: on Hugging Face model pages it detects model size and estimates VRAM requirements at FP16, INT8, and INT4 precision, then displays the cheapest compatible GPUs; on GitHub repositories it identifies model references and provides inline hardware estimates; and via a toolbar lookup it enables search for any model or GPU. Coverage breadth and daily refresh cadence are noted as current limitations compared to established trackers.