| Core question answered | Is this claim factually true? 0-100 credibility score per claim with sources linked | Was this text written by a human or an AI model? Confidence score per passage |
| Input format | Video and audio — YouTube, TikTok, Instagram Reels, podcasts — via link paste or the Chrome/Firefox extension while you watch | Pasted text and file uploads. GPTKit's site describes text analysis only; no video or audio detection is offered |
| Cross-referencing against live sources | Every extracted claim is checked against live web sources, with the evidence linked so you can verify the verdict yourself | Not offered — GPTKit analyzes the linguistic signature of the text itself rather than checking whether its content is accurate |
| AI-authorship detection of written text | Flags deepfake and AI-generated media signals in video/audio. Does not score whether a block of prose was AI-written | This is the entire product — six detection techniques, a stated accuracy of up to 93% from testing on 100,000+ human and ChatGPT responses, plus paraphrase detection |
| Language coverage | Any topic in any language — non-Latin-script videos route to a matching transcription and narration path | English only, per GPTKit's own documentation |
| API and bulk workflows | Consumer-facing: web app and browser extension. No public bulk-upload API | API access for integrating detection into your own platform, plus batch upload for analyzing many documents at once |
| Free tier | Free tier for link analysis and the extension. Basic $3.99/mo, Premium $8.99/mo | Free to use with a per-request character cap for guests; registering lifts the limit. Paid tiers add API and bulk features |