API

Authentication

Create a key under Settings → API keys and send it as a bearer token. Keys are shown once; revoke one there if it leaks. All requests and responses are JSON. Call the API from your server, not from a browser. Texts aren’t saved as documents, but on a free account (no plan or bought credits) they’re kept to train and improve our models, as described in the privacy policy.

curl https://qualifical.com/api/v1/credits \
  -H "Authorization: Bearer qk_…"

Passages

Texts are scored and rewritten a passage at a time, because detectors judge short text poorly: their scores become unreliable under about 600 characters. A passage is one paragraph (one line of the text). A paragraph under 600 characters, such as a heading, joins the paragraph after it; the last one joins the paragraph before. A paragraph over 2,500 characters is split at sentence boundaries into similar-sized parts. /rewrite takes passages by their text, so you send only the ones to rewrite.

Scores each passage of the text, then the whole text as the average of its passages weighted by their words. Costs a credit per word of the passages that were scored. Texts need at least 300 characters; results are more reliable above 600.

curl https://qualifical.com/api/v1/detect \
  -H "Authorization: Bearer qk_…" \
  -H "Content-Type: application/json" \
  -d '{"text": "When I first came to Oxford I got lost three times…"}'

{
  "aiScore": 0.49,
  "verdict": "mixed",
    "words": 162,
  "creditsCharged": 162,
  "attackDetected": false,
  "passages": [
    { "index": 0, "aiScore": 0.12, "words": 85, "start": 0, "end": 412, "text": "When I first came to Oxford…" },
    { "index": 1, "aiScore": 0.92, "words": 77, "start": 413, "end": 790, "text": "In today's rapidly evolving…" }
  ]
}

Rewrites passages so they read less like AI, usually the ones /detect scored 0.3 or higher. Send their text; nothing else of the text is needed. Each is rewritten paragraph by paragraph: headings and other short lines that don’t end a sentence stay as they are and aren’t charged. Costs seven credits per word sent for rewriting. Prose works best: tables, lists, and other markdown may not survive.

curl https://qualifical.com/api/v1/rewrite \
  -H "Authorization: Bearer qk_…" \
  -H "Content-Type: application/json" \
  -d '{
    "passages": ["In today's rapidly evolving…"],
    "tone": "casual"
  }'

{
  "passages": [
    { "index": 0, "rewritten": "Nobody tells you how much of the first term is admin…" }
  ],
  "creditsCharged": 539
}

A passage that comes back unchanged still costs its credits, because the model ran on it. Read rewrites before using them: one can shift the emphasis, change a term, or drop a detail.

Your plan, the credits you can spend, and what each action costs per word. Credits come from the free plan’s month (free), a plan’s period (subscription), and packs you bought (purchased), each null when you have none; expiresAt is when the first of them runs out. The ones that run out soonest are spent first. freeToday is the free credits used of today’s cap since 00:00 UTC, or null without free credits.

{
  "plan": "free",
  "isPaying": true,
  "creditsPerWord": { "detect": 1, "rewrite": 7 },
  "available": 33480,
  "free": { "available": 3480, "total": 5000, "expiresAt": "2026-11-01T00:00:00.000Z" },
  "subscription": null,
  "purchased": { "available": 30000, "total": 30000, "expiresAt": "2027-10-24T09:12:00.000Z" },
  "freeToday": { "used": 1520, "limit": 2500 }
}

Errors and limits

Errors share one shape. Branch on code; message is for people.

{ "error": { "code": "INSUFFICIENT_CREDITS", "message": "This needs 1,240 credits, more than you have left." } }