CostPerPrompt Tracks 232+ AI Models' Live Pricing
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- CostPerPrompt tracks live pricing for 232+ AI models with automatic refreshes, last updated 2026-08-02.
- The platform ships calculators that convert per-token rates into monthly cost estimates for chatbots, agents, and API workloads.
- AI providers bill per token (~¾ of a word) with separate input and output rates, where output typically runs 3–5× more expensive than input.
- Prompt caching cuts repeated input costs by up to 90%, a discount the tool flags as critical for chatbots that resend conversation history.
- Batch processing takes roughly 50% off when results can be delayed, and CostPerPrompt's calculators fold both discounts into their estimates rather than quoting raw token rates.
- The makers claim that most 'how much will this cost' articles omit caching and batching, producing estimates that run 2–3× too high or too low.
Why it matters: Engineering leads forecasting AI-feature budgets usually quote raw per-token rates, but prompt caching (up to 90% off repeat inputs) and batch discounts (~50% off) can swing monthly spend by 2–3×. CostPerPrompt bakes both into its calculators, giving teams a grounded baseline instead of the off-by-orders-of-magnitude estimates the site says dominate current 'AI cost' coverage.
