CostPerPrompt

CostPerPrompt is a free reference site that tracks live API pricing for large language models and turns those per token rates into monthly cost estimates. Its main table lists input, output and cached input prices per million tokens alongside context window size for more than 300 models from vendors including OpenAI, Anthropic, Google, Meta, Mistral, DeepSeek, xAI, Alibaba's Qwen, Moonshot, Cohere, Amazon and NVIDIA, filterable by vendor. It suits developers, product teams and finance owners who need a realistic budget before the first invoice arrives. A quick estimator on the home page takes requests per day and input and output tokens per request, and a fuller calculator models how APIs are actually billed, including growing chat history, cache hits, batch discounts and the gap between input and output rates. A separate GPU rental page compares on demand hourly rates for H100, A100, RTX 4090 and L40S cards across marketplaces, GPU clouds and hyperscalers, with a monthly cost calculator. The site is free to use and has no paid plans. It earns money through referral links, which are marked as such, and states that providers cannot pay to change their position in the price rankings. Prices are refreshed automatically from public listings, mainly the OpenRouter public model index, and cross checked against vendor pricing pages, with a refresh date shown on every page. The site presents its estimates as planning aids rather than quotes and is run by an independent developer.