LLM API price history

Price movement is most useful when combined with model fit, context window, output volume, provider region, and billing constraints. A cheaper input token rate can still lose when your workload has long outputs, heavy cached prompts, or a provider requirement that makes another route operationally simpler.

Start with the history log to spot fresh changes, open individual model pages for source URLs and confidence labels, then use the cost calculator to estimate monthly spend from your own token volume.

A logged change can mean a provider changed the input price, output price, cached-token rate, cache-write rate, or supporting metadata in the latest generated dataset. Treat each entry as a prompt to verify the provider source before changing production routing rules, especially when a row is marked low confidence or when the model family has multiple provider aliases.

For launch planning, the most important pattern is not a single cheap row. Look for stable provider coverage, recent observation dates, and a clear fallback route. If a model appears through OpenRouter and an official endpoint, compare both pages, then keep a note of the source timestamp used for your estimate.

Before a production release, pair this page with the sitemap and robots checks. The history page should stay crawlable, the playground should remain out of the sitemap, and every model URL in the current sitemap should return a clean static page. That keeps price updates discoverable without exposing experimental surfaces.