Estimate your monthly LLM API bill from tokens and traffic, include caching and batch discounts, and compare the same workload across popular models.
Prices are USD per 1M tokens. Pick a preset or enter your own.
Check your provider's terms. Cache and batch discounts vary.
Updates as you type.
Your selected model. “×” shows how many times more than the cheapest.
Estimates from list prices only (reviewed Nov 2026). Excludes tax, fine-tuning, tool calls, image or audio tokens and negotiated discounts. Verify with your provider.
Providers charge separately for input and output tokens, priced per million. The cost of a request is input tokens × input price + output tokens × output price, scaled to a million. Output tokens usually cost several times more than input tokens, so verbose answers drive the bill more than long prompts.
Two levers cut cost without changing the product: prompt caching (repeated context billed at a steep discount) and batch APIs (asynchronous work at a lower rate). Model choice matters most: the comparison above shows how far apart models can be on identical traffic.
Once you know your cost, the next step is pricing. Try the AI Pricing & Margin Calculator, and read about LLM cost tracking and attributing AI cost to customers.
Common causes: conversation history growing the input on every turn, retries, system prompts counted on every request, tool or function-call tokens, and reasoning tokens that are billed as output. Measure actual token usage per request and plug in the averages.
Providers can bill repeated prefix tokens (such as a long system prompt or document) at a discount when they are served from cache. Enter the share of input tokens you expect to be cached and the discount your provider offers.
The presets are representative list prices that change often. Edit the price fields to match your provider's current pricing page.
No. It runs in your browser, and the share link keeps your inputs in the URL fragment, which is not sent to our servers.