Become a Writer Today
Free tool · runs in your browser

AI Writing Cost Calculator

A 90,000-word novel is not 90,000 words of AI usage: drafting generates multiples of what you keep, and the model reads context on every call. Set your word count, drafting habits and timeline, and compare the true project cost of raw API access, Sudowrite tiers, and Novelcrafter with your own key.

The token maths

90,000 finished words at 3x drafting is 270,000 generated words. At the ~1.33 tokens-per-word heuristic that is 359,100 output tokens, plus an assumed 718,200 input tokens of context read along the way.

Raw API
$12.57

Claude Opus 4.8 pay-as-you-go for the whole project. No subscription, but you bring your own interface.

Novelcrafter + your key
$36.57

Hobbyist at $8/mo for 3 months, plus the same Claude Opus 4.8 API usage. Novelcrafter plans include no AI credits; AI runs on your own key.

Sudowrite
Hobby & Student$30.00
Professional$66.00
Max$132

Subscription only, for 3 months. Sudowrite meters usage in credits (225,000 to 2,000,000 per month by tier), and publishes no fixed credits-per-word rate, so credits are not converted to words here.

Raw API cost for this project, every model
ModelInput $/1MOutput $/1MProject cost
Claude Opus 4.8 Anthropic API$5$25$12.57
Claude Sonnet 5 Anthropic APIIntroductory pricing through August 31, 2026; $3 input and $15 output per million tokens from September 1, 2026.$2$10$5.03
Claude Haiku 4.5 Anthropic API$1$5$2.51
GPT-5.6 Sol OpenAI API$5$30$14.36
GPT-5.6 Terra OpenAI API$2.5$15$7.18
GPT-5.6 Luna OpenAI API$1$6$2.87
Every assumption in this estimate
  • ~1.33 tokens per English word. Heuristic, not a measurement: it inverts the published 1 token to roughly 0.75 words rule of thumb. Real tokenizers vary by model and content.
  • Drafting generates 3x your final wordage (your slider setting). Rerolls and cut scenes still cost tokens.
  • Each generated output token comes with 2x input tokens of context (your selector setting).
  • The project takes 3 months (your input), which sets the number of subscription billing cycles.
  • All prices verified on 2026-07-20 from the vendor pages listed under the tool. Prices change; check before you commit.

Runs entirely in your browser. Nothing you enter is sent anywhere.

Where the prices come from

Every rate in this tool was read from the vendor's live pricing page on 2026-07-20and is stored with its source URL and verification date. Nothing is estimated from memory, and rows that could not be verified on the day were left out rather than guessed.

API prices used (per 1 million tokens, verified 2026-07-20)

ModelInputOutputSource
Claude Opus 4.8Anthropic API$5$25pricing page
Claude Sonnet 5Anthropic APIIntroductory pricing through August 31, 2026; $3 input and $15 output per million tokens from September 1, 2026.$2$10pricing page
Claude Haiku 4.5Anthropic API$1$5pricing page
GPT-5.6 SolOpenAI API$5$30pricing page
GPT-5.6 TerraOpenAI API$2.5$15pricing page
GPT-5.6 LunaOpenAI API$1$6pricing page

Sudowrite plans (Hobby & Student $10/mo, Professional $22/mo, Max $44/mo) verified the same day atsudowrite.com/pricing; the page offers a yearly toggle advertising up to 50% off, so your invoice depends on the billing period. Novelcrafter plans (Scribe $4/mo, Hobbyist $8/mo, Artisan $14/mo, Specialist $20/mo) verified atnovelcrafter.com/pricing; no Novelcrafter plan includes AI credits, so AI usage bills to your own API key on top.

Publishing that AI-heavy draft on Amazon?

If an AI tool created actual text, images or translations in your book, KDP requires a disclosure, even after substantial edits. Walk your workflow through theKDP AI Disclosure Wizardbefore you hit publish, and check the AI stance of magazines and contests in theAI submission policies database.

Frequently asked questions

How many tokens is a word?

This calculator uses roughly 1.33 tokens per English word, which is the published 1 token to about 0.75 words rule of thumb turned around. It is a heuristic: real tokenizers differ by model, and dialogue, names and non-English text all shift the ratio.

Why multiply my word count before pricing it?

Because drafting with AI is not one clean pass. Rerolls, rejected passages, alternative versions and revision prompts all consume tokens, so the wordage you generate is a multiple of the wordage you keep. The slider makes that assumption explicit instead of hiding it; 1x means you keep every word, 5x means a heavy reroll habit.

Why is there an input-token setting?

APIs charge for what the model reads as well as what it writes. Every generation call resends your prompt, style notes, outline and often prior chapters as input tokens. The setting expresses that as a ratio of input to output tokens, and input is usually the cheaper of the two rates.

Which is cheaper: Sudowrite, Novelcrafter or the raw API?

It depends on your volume and your tolerance for setup. Raw API is pure usage pricing with no interface. Novelcrafter charges a small subscription for the interface and runs AI on your own API key, so its total is subscription plus the same API bill. Sudowrite bundles interface and usage into a credit allowance per tier. Run your own numbers above; the answer flips as word count and project length change.

Are these prices current?

Every price was read from the vendor's live pricing page on 2026-07-20, and each row in the tool's data file carries that verification date and source URL. AI pricing moves quickly, so confirm on the vendor's page before committing to a plan.

Newsletter

The writer's edit

One practical writing tip, a new tool, and a fresh contest deadline, every Thursday. No spam, unsubscribe anytime.