Google Veo 3: Pricing, Versions & Test Results
Google Veo 3.1 comes in Quality, Fast and Lite versions. We tested Veo 3.1 Fast against five competitors - here's what it costs and how it performed.

👁 114 reads
Veo 3 is Google’s family of video models that generate short clips with sound from text descriptions or images. The current generation is Veo 3.1 in three variants: full (Quality), Fast and Lite. Below we’ve gathered where Veo works today, what it costs at source in dollars, how it performed in our test using the same prompts as five competitors, and what it can’t do.
Quick Summary
Veo 3.1 Fast ranked second in our video model test (8.0/10) and delivered the best coffee-pouring shot. A 4-second 720p clip with sound cost $0.40 via API and was ready in 52-54 seconds. The cheapest access to Veo is through Google AI packages (from $6.22/month for 200 points in Flow), and the most convenient is through Gemini API or OpenRouter, where you pay per second. Limitations: clips are only 4, 6 or 8 seconds, aspect ratios are 16:9 or 9:16, and in the Gemini app the newer Gemini Omni replaces Veo.
New ranking: 10 video models tested with non-English speech - which handles multilingual dialogue without errors
01Veo 3.1 Pricing: Where It’s Available and Cost Per ClipVeo 3.1 pricing breakdown: Google AI packages from $6.22 to $254, Flow points system, cost per clip (Lite, Fast, Quality), and API rates per second.Read →
02AI Video Generator Ranking: 6 Models, Same PromptsWe tested Veo 3.1, Kling 3.0, Seedance 2.0, Hailuo 2.3, Grok and Wan 3.0 with identical prompts. Videos, costs, speed and scores compared.Read →
03Seedance 2.5: Versions, Pricing, Rankings and Test ResultsSeedance 2.5 ranks third in Artificial Analysis text-to-video, but our Polish-language tests exposed weak speech and text rendering.Read →
04Kling AI: How to Start, Check Cost and Review Video QualityKling AI generates video from text and images. We explain how to check your free credits, estimate real costs, and review clips before publishing.Read →
05Sora and Sora 2: OpenAI Shut Down Video GeneratorOpenAI's Sora video generator is gone. What happened, how to recover your videos, and what to use instead - including free alternatives.Read →Versions: Veo 3, 3.1, Fast and Lite
At the I/O conference on May 20, 2025, Google showed Veo 3 as its first model that generates image and sound in a single pass. On October 15, 2025, it was replaced by Veo 3.1 with better prompt adherence and control over first and last frames. Today the Gemini API pricing lists only 3.1 variants, so when we write “Veo 3,” you’re actually buying Veo 3.1.
| Variant | API Identifier | Resolution | Best For |
|---|---|---|---|
| Veo 3.1 (Quality) | veo-3.1-generate-preview | 720p, 1080p, 4K | shots for publication, highest quality |
| Veo 3.1 Fast | veo-3.1-fast-generate-preview | 720p, 1080p, 4K | daily production, 4x cheaper than Quality in 720p |
| Veo 3.1 Lite | veo-3.1-lite-generate-preview | 720p, 1080p | sketches, large batches, extending clips in Flow |
All three accept text and images for first and last frames, create clips of 4, 6 or 8 seconds, and generate sound together with video. On OpenRouter, Veo 3.1 has been available since March 2026, and Fast and Lite since April 2026.
Where Veo Is Available
- Google Flow - Google’s video studio for Google AI subscribers. All three Veo 3.1 variants work here alongside Gemini Omni, and Veo has exclusive access to clip extension. According to Google’s help, each account gets 50 points daily.
- Gemini App - according to Google’s video generation page, videos in the app are now made by Gemini Omni 1.1 Flash, which “will replace Veo in the Gemini app.” If you want Veo specifically, go to Flow or the API.
- Gemini API and Google AI Studio - pay per second, no free tier for Veo.
- OpenRouter and other aggregator platforms - same rates as Google, one account for many models. We used OpenRouter for our test.
Veo 3.1 Pricing at Source
Rates from the Gemini API pricing as of September 27, 2026, in USD excluding VAT, per second of finished video with sound. Google notes that you only pay for the generated clip.
| Variant | 720p | 1080p | 4K | 8s clip in 720p |
|---|---|---|---|---|
| Veo 3.1 (Quality) | $0.40 | $0.40 | $0.60 | $3.20 |
| Veo 3.1 Fast | $0.10 | $0.12 | $0.30 | $0.80 |
| Veo 3.1 Lite | $0.05 | $0.08 | not available | $0.40 |
OpenRouter also has cheaper rates without sound, for example Fast in 720p for $0.08 per second. In Google AI packages you pay with Flow points: Lite costs 10 points, Fast 20, and Quality 100 (8 seconds only); in the Ultra plan, Lite and Fast variants are half price. A full breakdown of packages is in the article Veo 3.1: pricing.
Our Test: Same Prompts as Five Competitors
On September 27, 2026, we gave Veo 3.1 Fast two prompts that also went to Wan 3.0, Kling 3.0, Seedance 2.0 Fast, Hailuo 2.3 and Grok Imagine Video. Parameters: 4 seconds (closest to 5s, since Veo doesn’t offer 5s), 720p, 16:9, with sound, one attempt. Full results are in the video AI generator ranking.
| Coffee | Tram | |
|---|---|---|
| Cost (usage.cost) | $0.40 | $0.40 |
| Time to file download | 52 s | 54 s |
| File | 1280×720, 4.0 s, with sound | 1280×720, 4.0 s, with sound |
| Speech in track (faster-whisper with VAD) | none | none |
| Rating | 8.5/10 | 7.5/10 |
The bill matches the pricing exactly: 4 seconds at $0.10 USD. Veo together with Grok was the fastest in the entire comparison, three times faster than Wan 3.0 and Hailuo. The audio track has no music or speech that we didn’t request; Seedance and Wan added a melody to the coffee.
Limitations
- Duration. One clip is 4, 6 or 8 seconds. Longer scenes are made by extending in Flow. Wan 3.0 and Seedance 2.5 create up to 30 seconds in one pass.
- Aspect ratios. Only 16:9 and 9:16, no square or 4:3.
- Motion speed. In our tram shot, movement was faster than in a real city. This is one shot, so we’re not generalizing, but it’s worth adding “slowly” to the prompt.
- No free API. Gemini API has no free tier for Veo. The only free option is 50 daily points in Flow, which don’t accumulate.
- Non-English dialogue. Google doesn’t publish a list of speech languages for Veo, and we didn’t test non-English dialogue in Veo. Wan 3.0 spoke a Polish line word-for-word in our test, which we described in the Wan AI hub.
Our take: Veo 3.1 Fast is a good default choice for short, realistic shots with sound, especially if you’re already paying for Google AI Pro for Gemini. For quick sketches Grok is cheaper, and for longer scenes Wan 3.0. Broader context is in the guide to video AI generators.
Frequently Asked Questions
How much does Veo 3 cost?
Through Gemini API, Veo 3.1 costs $0.40 per second, Fast $0.10, and Lite $0.05 in 720p, excluding VAT. In Google AI packages, access starts at approximately $6.22/month (AI Plus, 200 points in Flow).
Is Veo 3 free?
Partially. Google Flow gives each account 50 points daily, meaning for example 2 Fast clips or 5 Lite clips. Gemini API has no free tier for Veo.
What’s the difference between Veo 3 and Veo 3.1?
Veo 3.1 is the successor to Veo 3 with better prompt adherence and control over first and last frames. In the Gemini API pricing from September 2026, only 3.1 variants are listed: Quality, Fast and Lite.
Does Veo 3 work in the Gemini app?
According to Google, videos in the Gemini app are now made by Gemini Omni 1.1 Flash, which replaces Veo there. Veo 3.1 remains in Google Flow, AI Studio and the API.
Is Veo 3 available in the US and UK?
Yes. Google AI packages with Flow are sold in both markets, and the API works with accounts from these countries. We generated through OpenRouter without any workarounds.
Sources and verification date
- Gemini API pricing
- Google Flow help: points
- Google Flow help: models
- Google AI plans
- Google: video generation in Gemini
- OpenRouter: Veo 3.1 Fast
- TechCrunch: Veo 3.1 launch (Oct 15, 2025; Veo 3 announced at Google I/O May 20, 2025). Own test September 27
- 2026
- 2 clips
- $0.80. Verified September 27
- 2026
More on this from QUANTSLANT
benchmarks·Meta Muse Image: Price, Availability & Test ResultsMeta's Muse Image ranks 7th in Artificial Analysis at $0.01 per image. We tested it with five prompts and it scored 8.9/10, beating models 13x more expensive.

