What to compare before choosing an LLM API
The lowest headline input price is not always the cheapest production route. Output price, cache reads, batch discounts, context window, vendor availability, and billing minimums can change the effective cost.
- Compare input and output price per 1M tokens together.
- Check cache read and cache write pricing for repeated prompts.
- Confirm whether the vendor supports the exact model version you plan to use.
- Use the last verified date as a prompt to re-check vendor pricing pages.