AI Usage Cost Calculator
Estimate how much your AI API usage will cost. Pick a model, enter token counts, and see per-request and projected daily, monthly, and yearly costs.
$2.5 / $10 per 1M tokens (input / output)
Cost Per Request
1,500 tokens per request
Cost Projections
Based on 100 requests per day
What Is an AI Usage Cost Calculator?
An AI usage cost calculator helps you estimate how much you'll spend on large language model (LLM) API calls before you scale. By combining a model's per-token pricing with your expected input and output token counts, you can forecast costs per request and project daily, monthly, or annual spend.
How to Use This Calculator
Estimate your AI API costs in four simple steps:
- Select an AI model from the dropdown (GPT, Claude, Gemini, or enter custom pricing)
- Enter the average input and output tokens per API request
- Set your expected number of requests per day
- Review the per-request breakdown and daily, weekly, monthly, and yearly projections
How AI API Pricing Works
Most LLM providers charge separately for input tokens (your prompt, context, and system messages) and output tokens (the model's response). Prices are quoted per million tokens:
- Input tokens: Everything you send to the model, including prompts, conversation history, retrieved documents, and tool results
- Output tokens: The text the model generates in response, typically priced higher than input tokens
- Batch API: Some providers offer discounted rates (often 50% off) for non-real-time workloads
- Prompt caching: Repeated context can be billed at a reduced rate on supported models
Tips to Reduce AI API Costs
Once you know your baseline spend, these strategies can significantly cut costs:
- Use smaller models for simple tasks. GPT-4o Mini or Claude Haiku handle classification and extraction at a fraction of flagship model prices
- Trim context windows by removing unnecessary conversation history and limiting retrieved document chunks
- Cache repeated system prompts and static context where your provider supports prompt caching
- Set output token limits to prevent runaway generation on open-ended prompts
Frequently Asked Questions
How accurate are these cost estimates?
Estimates are based on published API list prices and your entered token counts. Actual bills may differ due to prompt caching discounts, batch pricing, rate limits, tool-use overhead, and pricing updates. Use this as a planning baseline, not an exact invoice.
What counts as an input vs output token?
Input tokens include everything you send: system prompts, user messages, conversation history, and any retrieved context. Output tokens are what the model generates. A rough rule of thumb: 1 token ≈ 4 characters in English, or about 750 words per 1,000 tokens.
Can I calculate costs for a custom or fine-tuned model?
Yes. Select 'Custom Model' and enter your provider's input and output prices per million tokens. This works for fine-tuned models, self-hosted inference endpoints, or any provider not in the preset list.
Where can I learn more about AI pricing and economics?
AI pricing changes frequently. For deeper coverage of model costs, coding agent economics, and infrastructure trends, check out Capital & Compute, a site focused on where money meets compute.