About
AI CostSaver is an independent catalog of AI inference providers and the models they serve. We track input/output prices per million tokens, context window sizes and endpoint availability, so engineers can pick the cheapest way to run any given model.
Where the data comes from
The catalog is seeded and refreshed from OpenRouter's public model catalog and, for providers where an admin has stored an API key, directly from each provider's own model-listing API (OpenAI-compatible, Anthropic, Gemini and Vertex dialects). Prices are synchronized automatically on a schedule and can always be re-run from the admin panel.
Why it matters
The same model is often served by a dozen providers at wildly different prices — a 10× difference for identical tokens is common. Comparing them manually is tedious; this catalog automates it.
Technology
Built with Python/FastAPI, PostgreSQL, SQLAlchemy and Tailwind CSS. Deployed with Docker. Machine-readable versions of this catalog are available at /llms.txt and /sitemap.xml.