AI features are cheap to prototype and easy to underestimate in production. Every request resends instructions and history, long conversations grow quietly, images get regenerated, and embeddings pile up in storage. The bill follows usage, not users.
These calculators break an AI feature into the parts you pay for. Start with tokens and a single request, scale it to a real feature, then add images, fine-tuning or retrieval if you use them. The LLM prices are checked against official pricing pages and dated on each page.
1. Tokens and a single request
Convert words to tokens, price one API call, and check how much of the context window your prompt, history and documents fill.
- Token to Word Count ConverterEstimate how many tokens a piece of text will use, or how many words a token budget allows, using the standard ~4 characters per token rule of thumb.
- AI API Cost CalculatorEstimate the cost of OpenAI, Claude and Gemini API calls per request and per month, from tokens, words, code lines or characters, with prompt caching and batch discounts.
- LLM Context Window CalculatorSee how a system prompt, chat history and documents fill an LLM's context window, how many turns fit before it overflows, and what each request costs.
2. A feature at real volume
Scale per-request costs to users and conversations, including the history that gets resent every turn.
- AI Feature Cost CalculatorEstimate the monthly API bill for a chatbot or AI feature from your users and conversations, including the cost of re-sending chat history.
- API Rate Limit ConverterConvert between requests per second, minute, and day, and get a safe sustained rate with headroom built in, not just the raw math.
3. Images, tuning and retrieval
Price usable images after retries, fine-tuning runs and the tuned model's running cost, and the storage and memory a vector database needs.
- AI Image Generation Cost CalculatorWork out what AI-generated images really cost per usable result, once retries, variations and storage are counted, and project the monthly bill.
- AI Fine-Tuning Cost CalculatorEstimate the cost of fine-tuning an LLM from your dataset size and epochs, then add the monthly cost of running the tuned model.
- Vector Database Storage CalculatorEstimate the storage and memory your embeddings need from the vector count, dimensions, precision and index type, including metadata and replicas.
- LLM VRAM Calculator: GPU Memory to Run a ModelEstimate how much GPU memory an open-weight LLM needs for inference: model weights at your chosen precision, the KV cache for your context length, and overhead.
Questions people ask
What drives LLM costs the most?
Input tokens resent on every request: long system prompts, attached documents and chat history. Output is priced higher per token, but input volume is usually larger. Caching and trimming history are the biggest levers.
How accurate are these estimates?
The arithmetic is exact; the uncertainty is in your inputs. Measure real token counts and conversation lengths from a pilot, then enter those figures for a reliable budget.
How often are the model prices updated?
They're checked against each provider's official pricing page, and the check date is shown on every page that uses them. AI pricing changes often, so confirm before committing to a budget.