Paste any text and our engine will analyze it for synthetic content, semantic coherence, and anomalous patterns — giving you a MarginT Score in under a second.
MarginT Score 70-100 — Authentic
The text shows natural language patterns, appropriate vocabulary distribution, and expected semantic coherence. Safe for AI training pipelines.
MarginT Score 40-69 — Suspicious
Some synthetic indicators detected. Review flagged sections. May be acceptable depending on your risk tolerance and use case.
MarginT Score 0-39 — Synthetic Detected
High probability of AI-generated content. Overused buzzwords, unnatural phrasing patterns, or anomalous statistical signatures detected. Not recommended for training.
Want to verify at scale? View our API plans
API
Integrate MarginT into your pipeline with a single API call. Get the same detailed analysis programmatically.
import margint
# Initialize with your API key
client = margint.Client("mt_live_xxxxxxxx")
# Verify a single text
result = client.verify("Your text here...")
# Result contains the same analysis
print(result["marginT_score"]) # 0.923
print(result["verdict"]) # "authentic"
print(result["processing_ms"]) # 42
import { MarginT } from '@margint/sdk';
const client = new MarginT('mt_live_xxxxxxxx');
const result = await client.verify('Your text here...');
console.log(result.marginT_score); // 0.923
console.log(result.verdict); // "authentic"
console.log(result.processing_ms); // 42
curl -X POST https://api.margint.co/v1/verify \\n -H "Authorization: Bearer mt_live_xxx" \\n -H "Content-Type: application/json" \\n -d '{"text": "Your text here..."}'
# Response:
{
"marginT_score": 0.923,
"verdict": "authentic",
"processing_ms": 42
}