February 16th, 2026
Improved
Fixed
For the last 6 weeks we’ve been working hard on improving Supadata performance, addressing YouTube platform changes and other quality issues.
Fixed intermittently missing createdAt and tags fields in YouTube metadata.
Fixed truncated YouTube channel descriptions (/v1/youtube/channel endpoint)
Added support for mobile subdomain (m.youtube.com) on /v1/metadata endpoint.
Allowing /live/ YouTube URLs on both /v1/transcript and /v1/metadata endpoints.
Fixed support of short TikTok URLs with /t/ in path.
Prioritized native captions over AI generated ones. Now if a video has both platform captions (eg. generated by YouTube) and generated by our AI, we will return platform captions first.
Fix returning all available languages when fetching a transcript.
Fix incorrect engagement values for Facebook Reels on /v1/metadata.
Improved reliability of long video (>1h) transcription.
Improved distinction between videos that don’t exist vs that are geo-blocked, age-restricted or require membership.
Added new plan (Supa) and an Auto Recharge multiplier that allows purchasing larger top-ups at once.
Created a new system for evaluating AI-generated transcripts with the long term goal of improving their quality (eg. better language recognition, avoiding hallucinations etc).
We’ve made some changes that may affect your apps using Supadata API:
Removed additionalData.transcriptLanguages field from /v1/metadata endpoint for YouTube videos (and analogically for /youtube/video). Please use /v1/transcript to get list of transcript language codes instead.
Shortened batch job results availability to 1h (previously 24h). Please make sure to retrieve job results promptly. After 1h attempts to get job status return a HTTP 404 error.