On this page
To cross-post Google Veo videos across YouTube Shorts, TikTok, Instagram Reels, Facebook Reels and LinkedIn without API rejections, export your vertical (9:16) clip at 1080p rather than 4K. According to Google Cloud's Veo 3.1 documentation, veo-3.1-generate-001 renders 4, 6 or 8-second video/mp4 files at 24 frames per second in 720p, 1080p or 4K — and while a 1080×1920 MP4 passes every network in our social media video specs table, a 4K vertical output (2160×3840) exceeds Instagram Reels' 1,920-pixel maximum width.
Google Veo 3.1 export specs vs. social platform API limits
Google Cloud's official documentation for the Veo 3.1 model family (veo-3.1-generate-001, veo-3.1-fast-generate-001 and the preview veo-3.1-lite-generate-001, updated October 22, 2026) defines exact output parameters for text-to-video, image-to-video and reference-to-video generation. Note that Google shut down the Veo 3.1 Preview endpoints in the consumer Gemini API on October 22, 2026, directing developers to the generally available Veo 3.1 models on Google Cloud's Gemini Enterprise Agent Platform or to Gemini Omni Flash.
We ran Veo 3.1's documented output profiles through the same validation rules used by our video spec checker and our complete guide to publishing AI-generated videos to social media to see where a raw Veo MP4 passes or fails each network's publishing API:
Property | Google Veo 3.1 output (Google Cloud docs) | YouTube Shorts & TikTok API | Instagram Reels & Facebook Reels API | Cross-posting verdict |
|---|---|---|---|---|
Container & MIME |
| MP4, MOV, WebM accepted | Instagram: MP4, MOV; Facebook Reels: MP4 only | Passes every network natively |
Duration | 4, 6, or 8 seconds (reference image-to-video locks to 8 seconds) | Shorts: up to 180 s; TikTok: up to 600 s (account limit applies) | Instagram: 3 s – 15 min; Facebook Reels: 3 – 90 s | All three durations (4 s, 6 s, 8 s) clear the 3-second minimum |
Aspect ratio |
| Shorts requires vertical or square (≤ 1:1); TikTok recommends 9:16 | 9:16 recommended on both | Use |
Resolution |
| TikTok: 360–4096 px per side; Shorts: no API pixel cap | Instagram: max width 1,920 px; Facebook: min 540×960 portrait |
|
Frame rate | 24 fps | TikTok: 23–60 fps; YouTube: not restricted | Instagram: 23–60 fps; Facebook Reels: 24–60 fps, fixed frame rate | Passes all networks (sits right on Facebook Reels' 24 fps floor) |
Audio track | Native sound generation supported on Veo 3.1 | Accepted | Instagram requires AAC audio up to 48 kHz | Verify AAC audio and fast-start before scheduling |
YouTube Dream Screen vs. standalone Veo MP4 exports
Creators encounter Google DeepMind's Veo in two very different workflows, and only one of them produces a portable file for multi-channel scheduling:
Inside YouTube Shorts (Dream Screen): At Made on YouTube, YouTube introduced Veo inside Dream Screen so creators can generate AI video backgrounds and standalone 6-second video clips directly inside the YouTube mobile creation flow. That clip lives inside YouTube's editor, carries an automatic YouTube disclosure label, and publishes only to your YouTube channel.
Standalone MP4 generation (Google Cloud / Google Flow): When you generate a clip through Google Flow or the Vertex AI / Gemini Enterprise Agent Platform API (
veo-3.1-generate-001orveo-3.1-fast-generate-001), you receive a downloadablevideo/mp4file. Because Veo 3.1 generates synchronized sound effects, ambient noise and dialogue natively, you can either publish a single 6- or 8-second loop or stitch multiple 8-second Veo shots into a 24- to 60-second vertical story.
If you export a 4K vertical clip from veo-3.1-generate-001 or stitch several 8-second Veo segments together, run this ffmpeg command to downscale to 1080×1920, lock a constant 30 fps H.264 stream with AAC 48 kHz stereo audio, and move the MP4 index (moov atom) to the front of the file as required by Instagram's API:
ffmpeg -i veo-input.mp4 -vf "scale=1080:1920:force_original_aspect_ratio=decrease,pad=1080:1920:(ow-iw)/2:(oh-ih)/2,setsar=1,fps=30" -c:v libx264 -profile:v high -pix_fmt yuv420p -crf 20 -maxrate 20M -bufsize 40M -g 60 -c:a aac -b:a 128k -ar 48000 -ac 2 -movflags +faststart veo-master-1080p.mp4
SynthID, C2PA and mandatory AI disclosure labels
Every video generated by Veo is watermarked by Google DeepMind's SynthID, which embeds an imperceptible digital watermark directly into the video pixels across every frame and into the audio waveform, alongside C2PA Content Credentials metadata. SynthID is engineered to survive cropping, filters and re-encoding.
However, embedded provenance metadata does not replace your obligation to declare realistic AI content when posting. Under the rules in our AI content labels guide and our interactive AI disclosure checker:
Photorealistic scenes (
realisticScene): If your Veo clip looks like real footage of people, cities, nature or events that never happened, a disclosure label is required on YouTube, TikTok, Instagram, Facebook and Threads, and under Article 50(4) of the EU AI Act for professional publishers.Native Veo 3.1 dialogue and sound (
aiVoice/aiMusic): Because Veo 3.1 generates human-sounding dialogue and music natively, Meta (Instagram, Facebook, Threads) requires an AI label for realistic-sounding AI audio even when the voice does not belong to a real person. YouTube and TikTok require a voice label only when the audio imitates a real person, though YouTube also requires disclosure when AI music is the main focus.Clearly stylized animation (
unrealistic): If your Veo prompt produces an obviously cartoon, claymation or fantasy clip with no realistic voice or music focus, none of the four regimes require an AI label.
YouTube API vs. TikTok and Meta API flags: When you mark a video as AI-generated in Uplonic, Uplonic sends the official API disclosure flags to TikTok, Instagram Reels and Facebook Reels automatically. YouTube's Data API v3 includes a status.containsSyntheticMedia boolean field, which Uplonic does not send yet — for realistic Veo Shorts, also enable the "Altered content" setting in YouTube Studio.
Importing and scheduling Google Veo clips with Uplonic
Uplonic does not generate AI videos; it is the validation and publishing layer between your AI video generator and your social channels. Once your Veo clip is rendered, you have two ways to move it into your publishing calendar:
Browser check and calendar upload: Drop the MP4 into our video spec checker and vertical safe-zone preview to confirm that captions or key visual subjects aren't hidden behind Shorts, TikTok or Reels buttons. Then upload once to the Uplonic calendar, toggle the AI content label, and write per-network captions — Uplonic trims each network's text to its character limit while preserving your trailing hashtags.
Automated URL import via REST or MCP: If your pipeline renders Veo 3.1 clips on Google Cloud and exports them to an HTTPS URL, Google Drive or Dropbox, pass the link to Uplonic's
POST /videos/importendpoint (or theimport_videosMCP tool). Each call accepts up to 10 video URLs (MP4, MOV or WebM up to 500 MB per file), strips temporary query tokens when storingVideo.sourceUrl, and supports anidempotencyKey(8–200 characters) plusrequireReview: truewhen clips must wait for human approval before scheduling.
curl --request POST "https://api.uplonic.com/videos/import" --header "Content-Type: application/json" --header "x-api-key: WORKSPACE_API_KEY" --data '{
"requireReview": true,
"items": [
{
"url": "https://storage.googleapis.com/your-bucket/veo-3-1-short-1080p.mp4",
"title": "Veo 3.1 vertical teaser (9:16, 1080p)",
"tags": ["google-veo", "ai-video", "shorts"],
"idempotencyKey": "veo-render-2026-10-25-clip-01"
}
]
}'
If you are also comparing other AI video models in your production workflow, see our companion guide on publishing HeyGen avatar videos across social networks.
Sources & methodology
Methodology: We verified Google Veo 3.1 technical specifications directly from Google Cloud's Gemini Enterprise Agent Platform documentation, Google AI for Developers deprecation notices, Google DeepMind's Veo and SynthID model pages, and the YouTube Official Blog on October 25, 2026. We then ran Veo's 1080p and 4K vertical parameters through Uplonic's checkVideoAllTargets and checkDisclosure modules.
Veo 3.1 model specifications (veo-3.1-generate-001, veo-3.1-fast-generate-001, veo-3.1-lite-generate-001) — Google Cloud Documentation, updated October 22, 2026, accessed October 25, 2026
Video generation in the Gemini API (Veo 3.1 Preview shutdown notice) — Google AI for Developers, updated October 22, 2026, accessed October 25, 2026
Veo 3.1 — Generate cinematic video with audio — Google DeepMind, accessed October 25, 2026
Google SynthID — Watermarking and identifying AI-generated content — Google DeepMind, accessed October 25, 2026
Made on YouTube: Google DeepMind Veo in YouTube Shorts Dream Screen — YouTube Official Blog, accessed October 25, 2026
Disclosing use of altered or synthetic content — YouTube Help, accessed October 25, 2026




