In April, Google officially launched Veo 4, successor to Veo 3.1, which had been released free for all Google users on April 2. The upgrade is substantial across three fronts that matter for content creators: duration, resolution, and narrative consistency.
What is new in Veo 4
Four concrete changes versus 3.1:
- Native storyboard — you write a multi-scene script and the model generates connected video, not isolated cuts
- Duration extended to 30 seconds per generation (was 8s in 3.1, 60s in Pro variants)
- 4K resolution (was 1080p)
- Zero-shot avatar creation — a single photo creates a consistent avatar across all subsequent scenes
The technical innovation nobody is talking about
The most interesting piece of Veo 4 is the Global Context Window. In previous models, if an object left the frame and came back, the model "forgot" its position/color/state. Result: cup in scene 1 was blue, the same cup in scene 3 turned green for no reason.
The Global Context Window keeps persistent memory of object position and state in the scene, even when they leave the frame. That is what separates "AI video" from "AI narrative". For long-form storytelling, it is critical.
Availability and pricing
Four tiers right now:
- Veo 3.1 (free for everyone with a Google account) via Google Vids or Flow
- Veo 3.1 Lite (cheaper API tier, launched April 7)
- Veo 3.1 Fast (reduced pricing for developers)
- Veo 4 (top tier, pricing starting at $249/month or pay-per-generation on Gemini API)
On the free tier, you can generate up to 12 videos per day on Veo 3.1. Enough for serious prototyping.
For creators: how to take advantage
Three workflows that became realistic with Veo 4:
- TikTok/Reels with continuous 30s narrative — used to require stitching 4 clips manually
- Cinematic product promo — built-in storyboard solves visual coherence
- Personal avatar for tutorials — one photo, infinite scenes with the same persona
Direct comparison: OpenAI Sora 2 still leads in pure cinematic aesthetic. Veo 4 leads in narrative consistency and personalization. For creator workflows, Veo 4 is more useful.
Sources
- DeepMind Veo official page
- Google Developers Blog (April 2026): Introducing Veo 3.1 and new creative capabilities in the Gemini API
- Google Blog: Veo 3.1 Lite — most cost-effective video generation model
- Veo 4 Release coverage (April 2026): Everything You Need to Know