| Released | 15 Oct 2025; Gemini API paid preview, Gemini app, Flow, Vertex AI |
|---|
| Max resolution | 1080p (8 seconds only); 720p default |
|---|
| Frame rate | 24 fps |
|---|
| Clip length | 4, 6 or 8 seconds |
|---|
| Videos per request | 1 |
|---|
| Aspect ratios | 16:9 and 9:16 (720p and 1080p) |
|---|
| Native audio | Yes, always on |
|---|
| Input types | Text; image; video (for extension) |
|---|
| Reference images | Up to 3 (8 seconds, 16:9 only) |
|---|
| First and last frame | Supported (interpolation); 8 seconds |
|---|
| Video extension | Extends Veo videos by 7 seconds per pass; input up to 141 seconds; 720p only |
|---|
| Request latency | Min 11 seconds; max 6 minutes at peak |
|---|
Gemini API video docs as archived 15 Oct 2025. 4K output arrived later (Jan 2026) and is not a launch spec. · source
| Spec | Value |
|---|---|
| Veo 3.1 Standard (video with audio) | $0.40 USD per second |
| Veo 3.1 Fast (video with audio) | $0.15 USD per second |
| Model codes | veo-3.1-generate-preview; veo-3.1-fast-generate-preview |
Paid tier only (no free tier). Google's launch post says Veo 3.1 is the same price as Veo 3 (developers.googleblog.com). Fast was cut to $0.10 per second at 720p and $0.12 at 1080p from 7 Apr 2026. · source
Estimate from the version's published specs and reviews, 0-5; scores are only comparable within one criteria year. Scored as launched (15 Oct 2025, Veo 3.1 and 3.1 Fast): 720p and 1080p output (4K upscaling came in Jan 2026, so not scored), 8-second clips that Extend chains beyond a minute (single generation stays 8 s), richer native audio, text plus up to 3 reference images, first and last frame, extension and insert (no video or audio references, so 3.5 not 4). No render time is published. Realism: Tom's Guide picks Sora 2 overall and flags repeated subjects and extra limbs; the AI Filmmaker likes the more human voices but sees more random cuts; Motion The Agency finds it more stable than Veo 3; Artificial Analysis's current arena ranks Veo 3.1 21st at Elo 963, a later board that mixes newer rivals, so the score is a middling 3.5.[1]
Veo 3.1 delivered a pretty good video, but I was disappointed by typical AI issues like repeating the dog and the flaw of the leash.
I noticed right away that the audio sounds more realistic, and the characters’ voices feel more human and less processed.
Veo 3.1 is more stable than its predecessor, with improved realism, sharper visuals, and more dynamic motion.
Veo 3.1 961 ±16
| Google Veo 3 review: Veo 3.1 tested in 2026 · Jul 10, 2025 | 0.60 |
|---|---|
| Testing the Limits of AI Video with Google Veo 3.1: Incremental Update or Major Shift? · Oct 18, 2025 | 0.50 |
| I tested Sora 2 vs. Veo 3.1 with 7 imaginative prompts — and there's a clear winner · Oct 24, 2025 | -0.20 |
Photos from readers, kept apart from the pin's sourced media. Rate what you ate or saw, and on a restaurant pin tag the dishes right on your photo.
The first entry is always the pin's source. Overall confidence is a weighted average of how firmly each reference supports the start and end times used above; a reference counts half as much for every 180 days older than the newest.
Google's post 'Introducing Veo 3.1 and advanced capabilities in Flow' and the Google Developers Blog[3] are both dated 15 Oct 2025; 9to5Google[6] reports 'Google today announced Veo 3.1'. Vertex AI listed general availability on 17 November 2025.
Arena lists Veo 3.1 at Elo 963 (rank 21) and Veo 3.1 Fast at 960, with launch dated January 2026 on the board.[1]
January 2026 update adding vertical (9:16) Ingredients to Video and upscaling to 1080p and 4K.[1]
Head-to-head test picking Sora 2 overall and Veo 3.1 for the jazz-club prompt, with repeated-subject and extra-limb errors from Veo.[1]
Veo 3.1 and Veo 3.1 Fast enter paid preview in the Gemini API; up to 3 reference images, scene extension for videos of a minute or more, first/last-frame transitions; same price as Veo 3.[1]
Something missing or wrong? Say it in your own words: a link that backs this pin up, a different start or end date and why, or a fact it lacks or gets wrong. The AI checks it against this pin's sources, searches for better ones, and adds any page that backs you up. The pin's own sources still count most. A picture that shows something else, or shows it badly, is looked at too, and moved down or replaced.