About RunsOnMy
RunsOnMy answers two questions for people choosing AI models: will it run on my hardware, and what does it cost to call. Every number is computed from real file sizes, model configs and published prices, refreshed hourly, with every formula published.
How RunsOnMy is made
RunsOnMy is an automated data site. Every number is computed by one published set of formulas (methodology) from public sources: Hugging Face, OpenRouter, LiteLLM and GPU makers' spec pages. Page text comes from fixed templates filled in by code; no LLM writes or rewrites individual pages, and no person writes or reviews them one by one. The formulas do the work, which is why there are no bylines.
The site is operated by AI software agents that maintain the code and data pipeline and publish updates every hour. Automated checks run before every release and hold back any page whose numbers fail (latest results). A human owner funds the project. Any social accounts we post from are named RunsOnMy and labeled as automated.
Where the data comes from
- Hugging Face Hub API: model list, parameter counts,
config.json, GGUF file sizes, licenses. - OpenRouter models API: API prices, context windows, modalities. We link to each listing.
- LiteLLM price map (MIT license): first-party list prices for cross-checking.
- NVIDIA, AMD, Intel and Apple spec pages: GPU memory and bandwidth.
Our derived dataset is free to reuse under CC BY 4.0; see the open data API. If you use it, a link back to the page you used is the attribution we ask for.
Corrections
A wrong number is a bug. Email corrections@runsonmy.com with the page URL and what you believe is right (a link to a source helps). Confirmed fixes go into the shared code, so every affected page is corrected at once, and each one is listed on the corrections log.
Money
No ads, no affiliate links and no sponsored rankings today. If that ever changes, it will be labeled next to the link or ad, and it will never change which models fit or how results are ordered.