New Nov 17: Enhanced model updated for Originality and Pangram
API Reference v1.3

AI Humanizer API Documentation

Humanize AI-generated text, detect AI authorship, and track usage through a simple REST API. Simple authentication, predictable pricing, 40+ languages.

Base URLhttps://wh2.oceanhubtool.com/__ext__/api.writehuman.ai
View Pricing
Starting at$0.14/ 1K words
Watch Video Guide

Quickstart

Get up and running in under a minute.

1

Get your API key

Sign up or log in at writehuman.ai, purchase an API subscription, and create your key from the dashboard.

2

Make your first request

Install an official SDK, or call the REST API directly:

npm install @writehuman/sdkpip install writehuman
// npm install @writehuman/sdk
import { WriteHuman } from "@writehuman/sdk";

const wh = new WriteHuman({ apiKey: "wh_live_your_api_key_here" });
const out = await wh.humanize("Artificial intelligence has revolutionized the way we approach content creation and digital communication.");
console.log(out.results[0]);
console.log(out.words_remaining.total, "words left");
3

See the response

Standard plans return a single result. Premium plans return 3 variations with human-likeness scores. On Premium, set enhanced to true to use the Enhanced model; the model field shows which one ran.

{
  "id": "hum_a1b2c3d4e5f6",
  "created": 1778533901580,
  "results": [
    "AI has changed how we create content and communicate online."
  ],
  "scores": null,
  "input_words": 14,
  "words_remaining": {
    "included": 124986,
    "topup": 0,
    "total": 124986
  },
  "model": "standard"
}

Video Guide

A step-by-step walkthrough of authentication and setup, followed by three real integrations: calling the API directly with Python, wiring it into a Lovable app, and using it with Claude.


Authentication

All API requests require a Bearer token in the Authorization header. You can create and manage API keys from your dashboard.

Authorization: Bearer wh_live_your_api_key_here

Keys use the prefix wh_live_ followed by a unique token. Keep your keys secret — treat them like passwords. You can create up to 5 keys and revoke any key at any time.


Plans & Limits

Choose a plan that fits your usage. All plans include monthly word allowances that reset each billing cycle.

Standard

Monthly words125,000
Max words per request2,000
Output variations1
Human-likeness scores
Rate limit40 req/min

Premium

Monthly words400,000 – 3,000,000
Max words per request4,000 – 6,000
Output variations3
Human-likeness scores
Rate limit120 – 160 req/min

Premium plans automatically return 3 output variations per request with human-likeness scores. Standard plans return a single result without scores. Usage is tracked by input-text word count. Words are deducted from your subscription first, then any top-up credits. Top-up packs never expire. Premium 400K allows 4,000 words per request and 120 requests per minute; 800K and up allow 6,000 and 160.


Errors

Errors return JSON with a code that says what went wrong and an error message for people. A 429, 503 or 504 comes with a Retry-After header: wait that many seconds, then retry.

StatuscodeMeaning
422invalid_requestThe body failed validation (missing text, too short or too long).
422forbidden_paramA parameter that is not available on the public API was sent.
422word_limitThe input is over the per-request word cap for your plan.
422moderation_blockedContent moderation refused the input (/v1/humanize only).
422freeze_tag_not_closedThe <freeze> tags in text are malformed (accounts with freezing only; see /v1/humanize). Nothing charged.
422freeze_tag_not_openedThe <freeze> tags in text are malformed (accounts with freezing only; see /v1/humanize). Nothing charged.
422freeze_tags_nestedThe <freeze> tags in text are malformed (accounts with freezing only; see /v1/humanize). Nothing charged.
422freeze_tag_emptyThe <freeze> tags in text are malformed (accounts with freezing only; see /v1/humanize). Nothing charged.
422nothing_to_rewriteAlmost all of the text is frozen, so there is nothing left to humanize (accounts with freezing only). Nothing charged.
401missing_api_keyNo usable Authorization: Bearer header.
401invalid_api_keyThe key is unknown or revoked.
402no_subscriptionThe key is valid but the account has no active API subscription.
402insufficient_balanceNot enough words left for this request; words_remaining is included.
403enhanced_requires_premiumenhanced: true was sent on a Standard plan. upgrade_url links to the plans. Nothing charged.
403enhanced_unavailableenhanced: true was sent while the enhanced model is switched off. Send the request without enhanced. Nothing charged.
403reverify_requiredThe account must be re-verified by email. The account owner signs in at writehuman.ai to continue; then the account's newest key created after the check began works within about a minute; every other key stays disabled until the check ends, so create a new key.
429rate_limitedOver your plan's requests per minute. Retry-After is the time to the next window.
429auth_locked_outToo many failed authentications from this address. Retry-After is the remaining lockout.
429queue_fullThe admission queue is full. Retry after Retry-After.
429queue_timeoutThe request waited in the admission queue past its deadline. Retry after Retry-After.
503at_capacityCluster-wide capacity shed. Retry after Retry-After.
503billing_unavailableThe billing store is unreachable; the request was refused rather than served unmetered. Retry after Retry-After.
503detector_unavailableThe detector, or the translation step before it, is unavailable (/v1/detect only). Retry after Retry-After.
504timeoutGeneration did not finish within the server deadline (/v1/humanize only). Nothing charged; retry after Retry-After.
500processing_failedGeneration raised an error.
500no_resultsGeneration finished without a usable result.
500safety_limitGeneration produced far more text than the input; discarded.
500billing_failedThe result was produced but the usage write failed part-way; included and/or top-up words may already have been charged. Check /v1/usage.

Endpoint Reference

Three REST endpoints, one base URL, and one Bearer key. All requests authenticate the same way.

POST/v1/humanize

https://wh2.oceanhubtool.com/__ext__/api.writehuman.ai/v1/humanize

Rewrite AI-generated text so it reads as naturally human. Billed by input word count, from your monthly allowance first, then top-up credits. The API accepts text, tone, language, enhanced, and freeze_citations. Premium plans can set enhanced: true for our enhanced model at the same word rate; the model field of the response says which model served it. Accounts with freezing enabled (English only, on request) can keep passages exactly as written by wrapping them in <freeze> and </freeze>, or keep citations with freeze_citations: true; the response then carries frozen_items.

Request Body — application/json

textrequired
string

The text to humanize.

>= 30 characters, <= 100,000 characters

toneoptional
string

Optional writing tone. Omit for default style.

Allowed: professional, academic, blog, casual, creative, scientific, technical

languageoptional
string | null

Target language (e.g. English, Spanish, Chinese (Mandarin)). Auto-detected if omitted.

enhancedoptional
boolean

Serve this request with our enhanced model, at the same word rate.

Default: false

Premium plans and above. On a Standard plan true returns 403 enhanced_requires_premium with an upgrade_url; while the enhanced model is switched off it returns 403 enhanced_unavailable.

freeze_citationsoptional
boolean

Keep the text's citations exactly as written: numbered references such as [1] or [2–4], bracketed keys such as [ref_3], author–year references such as (Smith et al., 2020), URLs and DOIs.

Default: false

Accounts with freezing enabled only (English only; ask support to enable it). Ignored on other accounts, where <freeze> tags are also treated as ordinary text. Can be combined with <freeze> tags in text; the tags are removed before anything else, never returned, never counted and never charged. Malformed tags return 422 with nothing charged. Not taken by the official SDKs yet; send it with the REST API.

200Text humanized successfully

Response Body — application/json

id
string

Unique identifier for this humanization request.

created
integer

Unix timestamp (milliseconds) when the request was received.

results
string[]

Humanized text variations, sorted by score (best first).

scores
(number | null)[] | null

Human-likeness score for each result (0–1, higher = more human).

Premium plans only. null on Standard plans.

input_words
integer

Number of words billed for this request.

words_remaining
object

Your remaining word balance after this request.

Fields: included (integer), topup (integer), total (integer)

model
string

Which model served the request.

Allowed: standard, enhanced

frozen_items
object[][]

Present only when the request used freezing. An array with one inner array per entry of results, in the same order; each inner array lists every frozen item in source order with its place in that result.

Each inner array holds objects { text (string), start (integer or null), end (integer or null) }; offsets in Unicode code points

An item that could not be kept has start and end set to null, and that part of the result is humanized like the rest.

Error Responses

401

Missing or invalid API key.

Invalid API key

{
  "error": "Invalid API key",
  "code": "invalid_api_key"
}

Missing Authorization header

{
  "error": "Missing or invalid Authorization header. Use: Bearer <api_key>",
  "code": "missing_api_key"
}
402

No active subscription or insufficient word balance.

No active subscription

{
  "error": "No active subscription",
  "code": "no_subscription"
}

Insufficient word balance

{
  "error": "Insufficient word balance",
  "code": "insufficient_balance",
  "words_remaining": {
    "included": 12,
    "topup": 0,
    "total": 12
  }
}
403

The request was refused before anything ran, and nothing was charged. The code says why.

enhanced on a Standard plan

{
  "error": "The enhanced parameter requires a Premium plan or above",
  "code": "enhanced_requires_premium",
  "upgrade_url": "https://wh2.oceanhubtool.com/api#pricing"
}

The enhanced model is switched off

{
  "error": "The enhanced model is not available right now",
  "code": "enhanced_unavailable"
}

Account must be re-verified

{
  "error": "This account must be re-verified by email. The account owner should sign in at writehuman.ai to continue.",
  "code": "reverify_required"
}
422

Invalid input, forbidden parameters, malformed <freeze> tags, over word limit, or content blocked by moderation.

Invalid request body

{
  "error": "Invalid request: ...",
  "code": "invalid_request"
}

model is not available on the public API

{
  "error": "The model parameter is not available",
  "code": "forbidden_param"
}

num_outputs is not available on the public API

{
  "error": "The num_outputs parameter is not available",
  "code": "forbidden_param"
}

sequence_id is not available on the public API

{
  "error": "The sequence_id parameter is not available",
  "code": "forbidden_param"
}

Over word limit

{
  "error": "Input exceeds per-request limit of 2000 words for your plan. Received ~2500 words.",
  "code": "word_limit"
}

Content blocked

{
  "error": "Content blocked: sexual",
  "code": "moderation_blocked"
}

A <freeze> tag is never closed

{
  "error": "A <freeze> tag is never closed. Add the matching </freeze>.",
  "code": "freeze_tag_not_closed"
}

A </freeze> tag has no opening tag

{
  "error": "A </freeze> tag has no <freeze> before it.",
  "code": "freeze_tag_not_opened"
}

<freeze> tags nest or overlap

{
  "error": "A <freeze> tag starts inside another one. Tags cannot nest or overlap.",
  "code": "freeze_tags_nested"
}

A <freeze> tag wraps no text

{
  "error": "A <freeze> tag wraps no text.",
  "code": "freeze_tag_empty"
}

Almost all of the text is frozen

{
  "error": "Almost all of the text is frozen, so there is nothing left to humanize.",
  "code": "nothing_to_rewrite"
}
429

Rate limit, queue capacity, or too many failed auth attempts.

Rate limited

{
  "error": "Rate limit exceeded for your plan (api_standard -> 40/min). api_standard: 40/min | api_premium: 120/min | api_premium 800K+: 160/min",
  "code": "rate_limited"
}

Too many failed auth attempts

{
  "error": "Too many failed authentication attempts",
  "code": "auth_locked_out"
}

Request queue full

{
  "error": "Too many queued requests. Please retry shortly.",
  "code": "queue_full"
}

Queue wait timeout

{
  "error": "Request waited too long for capacity. Please retry.",
  "code": "queue_timeout"
}
500

Processing failure. No words charged, except that billing_failed may have charged some of the request's words (included and/or top-up) before the usage write failed.

Generation failed

{
  "error": "Processing failed",
  "code": "processing_failed"
}

Generation returned nothing usable

{
  "error": "Processing returned no results",
  "code": "no_results"
}

Output exceeded the safety limit

{
  "error": "Output exceeded safety limit",
  "code": "safety_limit"
}

Billing update failed after generation

{
  "error": "Billing update failed",
  "code": "billing_failed"
}
503

Service at capacity, or billing temporarily unavailable. No words charged.

Service at capacity

{
  "error": "Service is at capacity. Please retry shortly.",
  "code": "at_capacity"
}

Billing temporarily unavailable

{
  "error": "Service temporarily unavailable, please retry shortly.",
  "code": "billing_unavailable"
}
504

The request did not finish within the server deadline (170 seconds from request entry; generation is stopped once 10 seconds remain so that billing is never cut part-way). No words charged; safe to retry.

{
  "error": "Processing did not finish within 160 seconds. Nothing was charged; please retry.",
  "code": "timeout"
}

GET/v1/usage

https://wh2.oceanhubtool.com/__ext__/api.writehuman.ai/v1/usage

Retrieve your current word usage, plan tier, and API version. Read-only — it never deducts words. Remaining balance = included_words − words_used_this_cycle + topup_balance.
200Usage returned

Response Body — application/json

included_words
integer

Words included in your plan each billing cycle.

words_used_this_cycle
integer

Words used so far in the current billing cycle.

topup_balance
integer

Remaining top-up words. Top-up packs never expire.

tier
string

Your plan: api_standard or api_premium.

version
string

Current API version.

{
  "included_words": 400000,
  "words_used_this_cycle": 18311,
  "topup_balance": 100000,
  "tier": "api_premium",
  "version": "v1.3"
}

Error Responses

401

Missing or invalid API key.

Invalid API key

{
  "error": "Invalid API key",
  "code": "invalid_api_key"
}

Missing Authorization header

{
  "error": "Missing or invalid Authorization header",
  "code": "missing_api_key"
}
402

No active subscription (HTTP 402).

{
  "error": "No active subscription",
  "code": "no_subscription"
}
403

The account owner must re-verify the account by email. Nothing is charged.

{
  "error": "This account must be re-verified by email. The account owner should sign in at writehuman.ai to continue.",
  "code": "reverify_required"
}
429

Too many failed auth attempts.

{
  "error": "Too many failed authentication attempts",
  "code": "auth_locked_out"
}

POST/v1/detect

https://wh2.oceanhubtool.com/__ext__/api.writehuman.ai/v1/detect

Score text for AI authorship using WriteHuman's detection model. Billed per input word (the original text), the same way as /v1/humanize. Non-English input is automatically translated to English before scoring. Set include_sentences to true for a per-sentence breakdown. Note: ai_probability runs 0–1 where 1 = AI — the opposite orientation of the humanize scores field.

Request Body — application/json

textrequired
string

The text to score for AI authorship.

>= 10 characters, <= 100,000 characters

languageoptional
string | null

Language of the text. Auto-detected if omitted; non-English text is translated to English before scoring.

include_sentencesoptional
boolean

Return a per-sentence AI-probability breakdown.

Default: false

200Text scored successfully

Response Body — application/json

scanId
string | null

Unique identifier for this detection scan.

version
string | null

Detector model version.

versionID
string | null

Detector build identifier.

ai_probability
number | null

Probability the text is AI-generated (0–1, where 1 = AI).

Opposite orientation from the /v1/humanize scores field, where higher = more human.

predicted_class
string | null

Overall label: ai, human, or mixed.

class_probabilities
object

Probability for each class.

Fields: ai, human, mixed

request_meta
object

Length of the scored text.

doc_length: { chars, tokens }

language
string

Detected (or supplied) language of the input.

sentences
array

Per-sentence breakdown, each { text, ai_prob }.

Present only when include_sentences is true.

{
  "scanId": "20260609OMIOISUZ",
  "version": "txtdetect-2026-05-14",
  "versionID": "V1.1.1",
  "ai_probability": 0.94,
  "predicted_class": "ai",
  "class_probabilities": { "ai": 0.94, "human": 0.04, "mixed": 0.02 },
  "request_meta": { "doc_length": { "chars": 642, "tokens": 98 } },
  "language": "English",
  "sentences": [
    { "text": "Artificial intelligence has rapidly transformed modern business.", "ai_prob": 0.97 },
    { "text": "Many teams now depend on it every single day.", "ai_prob": 0.61 }
  ]
}

Error Responses

401

Missing or invalid API key.

Invalid API key

{
  "error": "Invalid API key",
  "code": "invalid_api_key"
}

Missing Authorization header

{
  "error": "Missing or invalid Authorization header",
  "code": "missing_api_key"
}
402

No active subscription or insufficient word balance.

No active subscription

{
  "error": "No active subscription",
  "code": "no_subscription"
}

Insufficient word balance

{
  "error": "Insufficient word balance",
  "code": "insufficient_balance",
  "words_remaining": {
    "included": 12,
    "topup": 0,
    "total": 12
  }
}
403

The account owner must re-verify the account by email. Nothing is charged.

{
  "error": "This account must be re-verified by email. The account owner should sign in at writehuman.ai to continue.",
  "code": "reverify_required"
}
422

Invalid input or over the per-request word limit.

Invalid request body

{
  "error": "Invalid request: ...",
  "code": "invalid_request"
}

Over word limit

{
  "error": "Input exceeds per-request limit of 2000 words for your plan. Received ~2500 words.",
  "code": "word_limit"
}
429

Rate limit or too many failed auth attempts.

Rate limited

{
  "error": "Rate limit exceeded for your plan (api_standard -> 40/min). api_standard: 40/min | api_premium: 120/min | api_premium 800K+: 160/min",
  "code": "rate_limited"
}

Too many failed auth attempts

{
  "error": "Too many failed authentication attempts",
  "code": "auth_locked_out"
}
500

Billing update failure after a successful scan. Some of the request's words (included and/or top-up) may already have been charged before the write failed.

{
  "error": "Billing update failed",
  "code": "billing_failed"
}
503

Detector unavailable or at capacity, translation failed, or billing temporarily unavailable. No words charged.

Detector unavailable, at capacity, or translation failed

{
  "error": "Detection service unavailable. Please retry.",
  "code": "detector_unavailable"
}

Billing temporarily unavailable

{
  "error": "Service temporarily unavailable, please retry shortly.",
  "code": "billing_unavailable"
}

Code Examples

Full working examples in popular languages. Replace the API key with your own to get started.

Official SDKs: TypeScript / JavaScript @writehuman/sdk and Python writehuman. Both follow the documented retry policy: a 429, 503 or 504 is retried after the API's Retry-After, at most 2 retries and only when the wait is 60 s or less (both defaults); a longer Retry-After, such as an authentication lockout, is not waited for and the error surfaces at once with it as retry_after / retryAfter. A request that may have been charged is never resent.

// npm install @writehuman/sdk
import { WriteHuman } from "@writehuman/sdk";

const wh = new WriteHuman({ apiKey: "wh_live_your_api_key_here" });
const out = await wh.humanize("Artificial intelligence has revolutionized the way we approach content creation and digital communication.");
console.log(out.results[0]);
console.log(out.words_remaining.total, "words left");

Response

The response format depends on your plan.

{
  "id": "hum_a1b2c3d4e5f6",
  "created": 1778533901580,
  "results": [
    "AI has changed how we create content and communicate online."
  ],
  "scores": null,
  "input_words": 14,
  "words_remaining": {
    "included": 124986,
    "topup": 0,
    "total": 124986
  },
  "model": "standard"
}

Supported Languages

WriteHuman supports 40+ languages. Pass the language name as the language parameter, or omit it for automatic detection.

EnglishChinese (Mandarin)Chinese (Taiwan)FrenchSpanishVietnamesePortugueseJapaneseUkrainianSwedishGermanRussianItalianArabicHindiBengaliKoreanTurkishDutchPolishThaiIndonesianMalayPersianHebrewGreekCzechDanishFinnishNorwegianRomanianHungarianSlovakBulgarianCroatianSerbianSlovenianLithuanianLatvianEstonianUrduSwahiliTagalogAfrikaans

Common Use Cases


Support

By using the WriteHuman API you agree to our Terms of Service.

DIGI-TREASURE DIGI-TREASURE
Active
WriteHuman AI Active
Powered by Ocean Tool Hub