About CloudPrice.net

and AzurePrice.net

This website is an independent resource and is not affiliated with Microsoft, AWS, Google, OpenAI, Anthropic, or any other cloud or AI provider. It has been developed by individuals with the purpose of assisting people in finding information about cloud and AI pricing.

The primary objective of this website is to provide an easily accessible tool that displays VM/instance specifications and pricing for the major cloud platforms, alongside pricing, capabilities, and benchmark scores for AI and LLM models. All data presented on the website is sourced from publicly available APIs, price lists, and open datasets.

Data Sources & Credits

Nothing here is scraped from behind a paywall or a login. Every number traces back to a public price list, a provider API, or an open dataset. We read each source on a daily schedule, normalize it into a single schema, and keep every snapshot, which is what makes price history possible.

Cloud infrastructure

Pricing and specifications for cloud instances are sourced directly from the official public APIs of each provider:

AI and LLM Model Sources

The model catalog currently tracks roughly 3,000 models, of which about 1,400 carry live pricing across 75+ inference providers and 85+ regions. No single feed covers that surface, so we merge several classes of source and reconcile them onto one canonical model identity.

Model vendors (first-party price lists)

Where a vendor publishes its own prices, that is what we read. These are the authoritative numbers for each vendor's own API:

Cloud AI platforms

The same model often costs something different when bought through a cloud, and the price varies by region. These catalogs are read separately from the vendor lists above:

Gateways and inference marketplaces

Routers and marketplaces resell many vendors' models at their own rates, and are often the cheapest way to reach an open-weights model:

Catalogs and open data

Two open datasets give the catalog its long tail, and cover providers that publish no machine-readable price list of their own (Databricks, IBM watsonx, Oracle OCI, Replicate, fal.ai, Lambda, Baseten, Scaleway, Moonshot AI, MiniMax, Z.ai, AI21 Labs, Stability AI, ElevenLabs, Deepgram, Voyage AI, and others):

  • LiteLLM, whose model_prices_and_context_window.json is the broadest single feed we consume: open-source community pricing and context window data spanning 78 providers
  • Hugging Face Hub for open-weights model metadata, licences, and popularity signals (likes and downloads)

We publish the merged result back in the same shape: a LiteLLM-compatible price map lives at /api/v2/ai/litellm_model_prices.json, usable as a drop-in LITELLM_MODEL_COST_MAP_URL.

Benchmarks and performance
  • Artificial Analysis provides the quality and speed measurements: the Intelligence, Coding, and Math indices, plus MMLU-Pro, GPQA, HLE (Humanity's Last Exam), LiveCodeBench, SciCode, MATH-500, AIME and AIME 2025, IFBench, LCR (LeetCode Rating), TerminalBench Hard, and TAU2, alongside output throughput (tokens per second), time to first token, and Elo rating.
  • Hugging Face contributes additional inference throughput measurements for open-weights models.

All trademarks, product names, and logos belong to their respective owners and are used here only to identify the products being compared. Where a source disagrees with a vendor's own published price, the vendor wins. If you spot a number that looks wrong, please tell us.

Contact Us

Email: [email protected]