API Playground: run the humanizer API before you get a key

What you are looking at

The request panel is the literal call: POST https://tohuman.io/api/v1/humanizations/sync with a bearer token and a JSON body of content and intensity. The playground sends the same body to the same code path on an internal ToHuman account, so the key placeholder is the only thing that changes when you run it yourself. The response is rendered by the same serializer the API uses — nothing is reformatted for the page.

The response, field by field

• id and document_id — the humanization and the document it was saved under. GET /api/v1/humanizations/:id fetches it again later.
• status — completed, or failed with an error field when the model backend could not finish the run.
• output_content — the rewritten text.
• processing_time — seconds spent in the model, cold start included.
• words_remaining — your allowance for the month plus any purchased word credits, after this request. It is null here because the playground account is uncapped.
• scores — the model's own self-reported detectability, change_ratio and quality_fallbacks, explained in the questions below. Absent when the serving model does not report them.

Playground limits vs. the API

The playground takes up to 300 words per run and 3 runs per day per IP address; past that it answers 429 with a plain-English message rather than an API error. With your own key the synchronous endpoint accepts up to 2000 words per request, the asynchronous POST /api/v1/humanizations takes longer documents with a webhook, and the free plan includes 2,500 humanized words a month with no credit card. The full contract, error codes and rate limits are in the API reference; the step-by-step version is the humanize-AI-text API guide.

Questions developers ask before they get a key

Do I need an API key to use the playground?

No. The playground runs your text through an internal ToHuman account, so there is nothing to sign up for and no key to paste. It is capped at 3 runs per day per IP address and 300 words per run. A free API key lifts both limits: 2,500 humanized words a month, no credit card, and up to 2,000 words per synchronous request.

Is the playground response the same as the real API response?

Yes. The playground calls the same code path as POST /api/v1/humanizations/sync and renders the response with the same serializer, so the fields you see here — id, document_id, status, output_content, processing_time, words_remaining and scores — are the fields your own integration receives. The one difference is words_remaining: it is null here because the playground account is uncapped; with your key it is your remaining monthly allowance plus any purchased word credits.

What do detectability and change_ratio mean in the response?

They are the model's own self-reported scores, aggregated over paragraphs. detectability is the pipeline's mean AI-probability estimate for the text it shipped, from 0 to 1, where lower reads more human. change_ratio is the mean normalized edit distance from your input — 0.40 means roughly 40% of the text was reworked. quality_fallbacks counts paragraphs where no candidate met the quality bar and the least-detectable one shipped anyway; a re-run at a higher intensity is the usual fix. None of these is a third-party detector score, and no tool can promise a score from a closed classifier.