Estimate how much GPT-4o, Claude and Gemini will cost per day, month and year. Set input/output tokens per request and how many requests you send — all in your browser.
OpenAI, Anthropic and Google charge by the token, not by the request. Every prompt you send (input) and every word the model writes back (output) is billed separately. Output is almost always more expensive than input — often 3–5× — because generating text costs more compute than reading it.
Monthly cost is a straight multiplication:
Example: 1,500 input tokens + 400 output tokens, 10,000 requests/month on GPT-4o ($2.50 / $10.00 per 1M tokens) ≈ $37.50 input + $40.00 output = $77.50 / month.
Built-in rates are public list prices for standard (uncached) API input and output. Providers change them. Override the $/1M fields if your contract, region or cache discount is different. ChatGPT Plus / Claude Pro subscriptions are flat monthly fees and are not what this calculator models — this is for the developer APIs.
All calculations run instantly in your browser. Your usage numbers are never sent to any server and never stored.
Monthly cost = input tokens × requests × input price per million, plus output tokens × requests × output price per million. This tool assumes 30 days in a month. Output tokens (the model's reply) are usually several times more expensive than input tokens.
Input tokens are everything you send: system prompt, user message and any attached context. Output tokens are the model's reply. You pay for both. A long system prompt is billed on every request, even if the user message is short.
No. Those are flat consumer subscriptions. This calculator is for the developer APIs (OpenAI, Anthropic, Google), which bill per million tokens. If you only use the chat apps, you pay the subscription — not these per-token rates.
List prices change, and you may have batch, cached or committed-use discounts. The defaults are public standard rates. Type your real $/1M (or equivalent in another currency after converting) to match your invoice.
For most high-volume work, GPT-4o mini, Gemini Flash and Claude Haiku are an order of magnitude cheaper than flagship models. Use a small model by default and reserve GPT-4o, Sonnet or Opus for tasks that actually need them.
Yes. Everything runs locally in your browser. Nothing is uploaded, logged or stored.