Know the AI margin before you ship.

What this calculator covers

In one sentence

It turns a monthly usage estimate into a per-model API bill so you can compare OpenAI, Anthropic, and Google list prices before you commit to a model.

What you enter

Monthly volume, either as active users times requests per user or as a single request count, plus the average input and output tokens per request and the models you want to compare. Paying users and price per paying user are optional and only affect the revenue share figure.

What you get

For each selected model: the estimated monthly cost, the cost per request, the cost per active user while you estimate volume by active users, and AI cost as a share of monthly revenue once you add revenue. Models are ordered from the cheapest estimate for your scenario.

Worked example

A support assistant has 2,000 active users who each send 15 requests a month, so 30,000 requests. Each request averages 1,200 input tokens and 350 output tokens.

On GPT-6 Luna at $0.10 input and $0.50 output per 1M tokens, one request costs (1,200 × 0.1 + 350 × 0.5) ÷ 1,000,000 = $0.000295, and the month comes to $8.85. The same month on each model, at the list prices last checked 2026-09-23:

• GPT-6 Luna (OpenAI): $8.85 a month, $0.000295 per request
• Gemini 3.5 Flash-Lite (Google): $37.05 a month, $0.001235 per request
• Gemini 3.8 Flash (Google): $66.38 a month, $0.002213 per request. Price through Dec 31, 2026. From Jan 1, 2027: $1.50 input / $7.50 output per 1M tokens.
• Claude Haiku 4.5 (Anthropic): $88.50 a month, $0.00295 per request
• GPT-6 Sol (OpenAI): $177.00 a month, $0.0059 per request
• Claude Sonnet 5 (Anthropic): $177.00 a month, $0.0059 per request

Assumptions and exclusions

• Cost per request is (input tokens × input rate + output tokens × output rate) ÷ 1,000,000, and the monthly figure is that cost times your monthly request count.
• Token counts are the averages you type in. The calculator never reads or tokenizes your prompts, so it cannot count tokens for you.
• The same token counts are applied to every model, but tokenizers differ, so the same text can come out as a different number of tokens on each model.
• Standard text tokens at public list prices in USD only. Prompt caching, batch discounts, tool and function calling surcharges, image, audio and video rates, fine-tuning, taxes, and negotiated or committed-use tiers are all excluded.
• Rates are the base context tier. Long-prompt surcharges, such as OpenAI GPT-6 pricing above 272K input tokens, are not applied.
• Every month is treated as identical. There is no growth curve, seasonality, retry volume, or rate-limit effect.

Price source and date

Published list prices, last checked 2026-09-23. Providers change prices without notice, so confirm against the source before you rely on a number.

• OpenAI official pricing
• Anthropic official pricing
• Google official pricing

Privacy

• The whole calculation runs in your browser. tiny-tools has no backend, so your figures are never sent anywhere to be computed, and they are gone when you reload the page.
• Anonymous product analytics go to PostHog: that the calculator was viewed, started and completed, and the hostname of the site that linked you here. Your inputs, results and model selection are not included in those events.
• Your IP address reaches PostHog, which is configured to derive a country and time zone from it and then discard it rather than store it on the events.
• Session replay is part of the same analytics setup, so your on-screen session can be recorded. Input masking is applied by a PostHog project setting rather than by anything on this page, and values rendered on the page, including the estimate, can appear in a recording either way.
• PostHog stores an anonymous visitor identifier in this browser. Nothing else about your scenario is saved on this device.