AI token cost calculator — estimate your LLM API bill

Understanding LLM API pricing

Large language model APIs charge per token — a unit roughly equal to three quarters of an English word. Every request has two billable parts: the input (your prompt, system instructions and any context you attach) and the output (the text the model generates). Providers price these separately, and output tokens are typically 2–5× more expensive because generation is more compute-intensive than reading.

For a typical chatbot handling 10,000 conversations a month with average prompts of 500 tokens and responses of 300 tokens, the monthly bill ranges from under $2 on a lightweight model like GPT-4o mini to over $200 on a frontier model like Claude 3.5 Opus. That 100× difference is why picking the right model for each task matters far more than optimising prompt length.

GPT vs Claude vs Gemini — which API to choose

OpenAI' s GPT-4o family offers a broad range from the budget-friendly 4o mini to the reasoning-heavy o1 series. Anthropic' s Claude excels at long-context tasks, supporting up to 200K tokens in a single call — ideal for document analysis and code review. Google' s Gemini 2.0 is competitively priced and tightly integrated with Google Cloud, making it a natural choice if your infrastructure is already there. None of them is universally cheapest or best: the right pick depends on your latency, quality and budget trade-offs.

Frequently asked questions

How is the token cost calculated?

Cost = (input tokens ÷ 1,000,000 × input price) + (output tokens ÷ 1,000,000 × output price). Providers bill separately for input (prompt) and output (completion) tokens, which is why both prices are shown.

How accurate is the token estimate from pasted text?

The text mode uses a heuristic (a blend of character and word counts) that lands within roughly 10–20% for English. For exact billing, switch to token-count mode and paste the token totals your API response reports.

Are the model prices up to date?

The defaults were reviewed in July 2026, but providers change pricing often. Every price field is editable, so you can enter the current numbers from the provider's pricing page.

Is this calculator free?

Yes. Estimating per-call and monthly cost for any single model is free, runs entirely in your browser, and needs no sign-up. Toolizto Pro ($9.99/month or $79.99/year, covering every Toolizto tool) adds a side-by-side comparison of every model for your exact usage, a cheapest-model highlight, CSV export, and removes ads.

What is a token in the context of LLMs?

A token is a chunk of text — roughly three quarters of a word in English. The sentence 'Hello, how are you?' is about 6 tokens. LLM providers charge per token for both the prompt you send (input) and the text the model generates (output), so understanding token counts is essential for budgeting API usage.

How many tokens is 1,000 words?

In English, 1,000 words is approximately 1,300–1,400 tokens. The ratio varies by language and content type: code tends to use more tokens per word because of special characters, while simple prose is more predictable. Use the text-paste mode in this calculator for a quick estimate.

Which is the cheapest LLM API model right now?

As of mid-2026, the most affordable options for general-purpose use are GPT-4o mini (OpenAI), Claude 3.5 Haiku (Anthropic) and Gemini 2.0 Flash (Google) — all under $1 per million input tokens. The best choice depends on your quality requirements: cheaper models trade some reasoning depth for lower cost. Use the Pro comparison view to see every model side by side for your exact usage.

How can I reduce my LLM API costs?

Three practical strategies: shorten your prompts by removing unnecessary context (fewer input tokens), use a smaller model for routine tasks and reserve expensive models for complex reasoning, and cache repeated prompts so you do not pay twice for the same question. Many providers also offer batch-processing discounts for non-real-time workloads.