Price a prompt across model providers.
How to use Token Cost Estimator
Paste your prompt. Type or paste your prompt into Prompt text. The input token count appears under the box.
Set the expected output length. Enter a number in Estimated output tokens.
Set your request volume. Enter Requests per day (for daily cost) to scale the per-request cost.
Compare models. Read the Cost per request, by model table. Rows are sorted from cheapest to most expensive, with a per-request and a per-day column.
Limits and privacy
- Prices come from a hand-maintained table in the tool's code, labelled as a 2025-Q4 snapshot. They may be out of date, so check your provider's current pricing.
- The table covers ten models: GPT-4o, GPT-4o mini, o1, Claude Opus 4, Claude Sonnet 4, Claude Haiku 3.5, Gemini 1.5 Pro, Gemini 1.5 Flash, Llama 3.1 70B and Mixtral 8x7B. Llama and Mixtral prices are for hosted versions.
- Input tokens are counted with Work Easy's own illustrative BPE tokenizer, not each provider's real tokenizer, so the count is an approximation.
- Cost is input tokens times the input rate plus your estimated output tokens times the output rate. Caching, batch discounts, tool use and image or audio tokens are not modeled.
- Calculations run in your browser. Nothing is uploaded.
Frequently asked questions
How is the cost calculated?
Input tokens divided by one million times the model's input price, plus output tokens divided by one million times its output price. Daily cost multiplies that by your requests per day.
Are the prices up to date?
Not necessarily. They are a snapshot stored in the tool's code and labelled 2025-Q4, with no live feed. Verify current rates with the provider.
Is the token count exact?
No. It uses a small illustrative tokenizer, so counts can differ from what each provider bills. Treat the result as an estimate.
Does it include prompt caching or batch discounts?
No. It uses a single input price and a single output price per model.