CostPerPrompt

About CostPerPrompt

CostPerPrompt is a live reference for what AI actually costs to run, combining an automatically refreshed price table for 232-plus models with a set of calculators that turn per-token rates into the number you actually need — the monthly bill for a specific workload. The price table covers input, output and context window per model across every major vendor and is stamped with its refresh date, which matters in a market where published rates change several times a year and most comparison pages quietly go stale. The calculators are where it earns its place: an API cost calculator compares one workload across all 232 models with caching and batch discounts applied; a chatbot calculator simulates real conversations with growing history, resent context and cache hits rather than assuming a flat per-message cost; an agent calculator models multi-step loops, tool schemas and retries and shows why agent workloads run ten to thirty times more expensive than a naive estimate; and a RAG calculator prices indexing, retrieval and generation separately. There are also voice AI calculators covering the three-part STT plus LLM plus TTS bill per minute and per call, image API pricing across three market tiers, GPU rental prices for H100, A100 and RTX 4090 across ten providers where the same chip shows a fivefold spread, and a browser-local token counter. Cost-cutting guides cover prompt caching and the Batch API.

Key Features

CostPerPrompt Pros & Cons

Who Is CostPerPrompt Best For?

Tags