AI Humanizer API for Org-Wide LLM Output
Humanize LLM output across every department with Bearer token auth, up to 5 revocable keys, and 40+ auto-detected languages. Word-based plans start near $0.17 per 1,000 words, with read-only usage tracking so spend stays predictable as you scale.
Word-based pricing from $0.17 per 1K words. Bearer auth. 40+ languages.
- Bearer-token auth, keys prefixed wh_live_
- Up to 5 keys, revocable at any time
- Daily usage and word spend in your dashboard
- 40+ languages for global teams
Built for enterprise teams
Process LLM-generated content across the organization with reliability, control, and predictable cost.
Access you can control and revoke
Authenticate every call with a Bearer token (wh_live_ prefix). Issue up to 5 keys per account, name one per team or environment, and revoke any one of them the moment a contractor rolls off or a key leaks.
Predictable, word-based budgeting
Plans bill by input words and reset each cycle: Standard at $29 for 125,000 words, Premium at $69 for 400,000 words. Effective cost starts near $0.17 per 1,000 words, and 125,000-word top-up packs never expire when a team runs hot.
Usage you can monitor
Your dashboard shows daily request volume and word usage, and the read-only GET /v1/usage endpoint reports per-cycle consumption without ever incurring a charge. Finance and platform owners get an independent line of sight into spend.
Global, multi-format coverage
40+ languages are auto-detected from the input text, and seven tones (professional, technical, scientific, academic, blog, casual, creative) match the format. One endpoint serves comms, docs, reports, and customer-facing copy across the org.
Control and visibility built in
Predictable budgeting, revocable keys, and read-only usage tracking, so platform and IT teams can roll it out with confidence.
Bearer token authentication
Every call carries an Authorization: Bearer wh_live_... header. Keys are tied to your account, easy to name per team or environment, and never embedded in client-side code.
Revocable keys, up to 5
Issue up to 5 keys per account and revoke any of them instantly when a project ends or a credential is exposed. The MCP connector uses OAuth 2.1 with PKCE instead, so there are no shared keys to leak.
Usage visibility, no billing impact
Your dashboard shows daily request volume and word usage, and the read-only GET /v1/usage endpoint reports per-cycle consumption without billing, giving platform and finance owners an independent line of sight into spend.
Global language coverage
40+ languages are auto-detected from the input text, so a single endpoint serves regional comms, docs, and customer-facing copy across the whole organization.
One API, every team that ships LLM output
Internal comms
Org-wide announcements and leadership updates that read naturally
Documentation
Product docs and knowledge-base articles in a consistent technical tone
Customer support
Help-center copy and macros that sound human across 40+ languages
Reports
Internal and stakeholder reports refined from LLM first drafts
Drop humanization into your content pipeline
Generate a wh_live_ Bearer key for each business unit or environment from your account, with up to 5 active keys and revocation on demand. Send drafts from your internal tools to POST /v1/humanize with a tone and optional language, and return human-quality text that passes AI-detection checks. Poll GET /v1/usage to watch word consumption per cycle, set internal alerts before limits, and reconcile against word-based plans without any billing impact.
curl -X POST https://api.writehuman.ai/v1/humanize \
-H "Authorization: Bearer wh_live_your_team_key" \
-H "Content-Type: application/json" \
-d '{
"text": "Our Q3 internal update covers headcount, roadmap shifts, and the revised support SLA.",
"tone": "professional",
"language": "english"
}'MCP for assistant-native internal workflows
Beyond the REST API, the WriteHuman MCP connector lets internal assistants humanize and score text without handling raw keys. It authenticates with OAuth 2.1 and PKCE, so there are no shared secrets to distribute or rotate across an organization, and it is included with Pro, Ultra, and API plans. Connect it at /mcp.
Explore the MCP connector- OAuth 2.1 with PKCE, no shared API keys to manage across teams
- Works with Claude, Cursor, Codex, and any MCP client
- Tools: humanize_text, detect_ai_text, and get_account
Simple, word-based pricing
Pay for the words you send. Allowances reset every billing cycle, and top-up packs never expire.
Standard
$29 / mo- 125,000 words / month
- 40 requests / min
- Single humanized result per request
Premium
$69 / mo- 400,000 words / month
- 120 requests / min
- 3 variations with human-likeness scores
Frequently asked questions
How do we control and revoke API access across teams?
Every request authenticates with a Bearer token using the wh_live_ prefix. You can issue up to 5 keys per account, assign them to different teams or environments, and revoke any key at any time. For assistant-native workflows, the MCP connector uses OAuth 2.1 with PKCE, so there are no shared secrets to rotate.
Is spend predictable when many teams share one account?
Yes. Plans bill by input words and reset each cycle: Standard is $29 for 125,000 words and Premium is $69 for 400,000 words, with an effective rate near $0.17 per 1,000 words. Need more headroom? 125,000-word top-up packs never expire, and the read-only GET /v1/usage endpoint lets you monitor consumption without any billing impact.
Can we monitor usage across teams?
Yes. Your dashboard shows daily request volume and word usage, and the read-only GET /v1/usage endpoint reports per-cycle word consumption without any billing impact. Platform and finance owners get an independent line of sight into spend, separate from the teams generating the calls.
Does the API handle our global, multi-format content?
It does. The humanize endpoint auto-detects 40+ languages from the input text and accepts seven tones including professional, technical, scientific, and academic. That covers internal comms, documentation, reports, and customer-facing copy from one endpoint, with an average response near 20 seconds.
Can the API also detect AI-written text?
Yes. POST /v1/detect returns an ai_probability from 0 to 1 (where 1 means AI) and a predicted_class of ai, human, or mixed, with an optional per-sentence breakdown. Governance teams use it to screen output before publishing or to spot-check content at volume. See /ai-detector for the hosted tool.
What are the request limits and input sizes?
The humanize endpoint accepts text from 30 to 100,000 characters and bills by input words, with a per-request input limit up to 2,000 words. Rate limits are 40 requests per minute on Standard and 120 requests per minute on Premium. Premium also returns 3 scored variations per request.
Bring controlled humanization to every team
Provision your first Bearer key, route LLM output through one endpoint, and watch spend with read-only usage. Review plans and limits at /api#pricing, read the auth and endpoint reference at /api/docs, or connect the MCP server at /mcp for assistant-native workflows.