クイックアンサー
Veo 3.1 is Google's current Veo generation on the Gemini API, sold in three tiers: Standard ($0.40/s at 720p and 1080p, $0.60/s at 4K), Fast ($0.10–$0.30/s) and Lite ($0.05–$0.08/s, no 4K). Audio is included and there is no free tier.
Clips run 4, 6 or 8 seconds; 1080p and 4K require the 8-second setting. Scene extension adds 7 seconds per step, up to 20 steps, reaching roughly 148 seconds from a Veo-generated input.
Veo 3.1-specific controls are first–last frame interpolation and up to 3 ingredient reference images (neither on Lite). Google now positions Gemini Omni Flash as its cheaper default video model (~$0.10/s effective at 720p), with Veo kept for these controls.
Get an AI Studio API key and call veo-3.1-generate-preview, -fast- or -lite tiers, billed by the output second.
Veo pricingCreator-facing Veo workflows; plan and region access depend on your Google account.
Open FlowVeo weights are closed, outputs carry SynthID watermarking, and generated videos are stored for 2 days.
See accessVeo 3.1 is hosted-only — every mode runs through Google's API or apps and bills per output second.
T2VGenerate a clip from a scene prompt; audio is always generated with the video.
text → video + audio
I2VAnimate a starting image with a prompt; pair the image with lastFrame for first–last interpolation.
image + text → video
R2VUp to 3 reference images (people, characters, products) steer the clip. Not available on Lite.
images + text → video
EXTExtend a Veo-generated video by 7 seconds per step, up to 20 steps (~148s total), at 720p. Not on Lite.
video → video
FLFInterpolate from a start image to a lastFrame for exact transitions; both frames are required together.
2 images + text → video
Gemini API paid-tier rates per output second, checked against Google's pricing page on 2026-09-28 (unchanged from the prior check). Audio is included, and Google states you are billed only when a video actually generates.
| Tier | Price per output second | What it covers | Source |
|---|---|---|---|
| Veo 3.1 Standard | $0.40 (720p · 1080p) · $0.60 (4K) | Highest quality; every mode including ingredients and extension | 公式情報源 ↗ |
| Veo 3.1 Fast | $0.10 (720p) · $0.12 (1080p) · $0.30 (4K) | Faster, cheaper iterations | 公式情報源 ↗ |
| Veo 3.1 Lite | $0.05 (720p) · $0.08 (1080p) | Lowest cost; no 4K, no ingredients, no extension | 公式情報源 ↗ |
例: A ceramicist shapes a wet clay bowl in a sunlit studio, slow push-in from hands to face, wheel hum and soft room tone, documentary realism
Veo 3.1's strongest published results come from third-party human preference tests: two models received the same prompt, human raters picked the better video, and these clips won by wide margins. The prompt text is shown exactly as submitted.
Aerial view of synchronized swimmers performing intricate patterns in a crystal-clear lake, their movements fluid and graceful under the soft glow of the morning sun.
Preferred over Genmo Mochi 1 by 95% of raters on this shared prompt.
See the full Google Veo 3.1 vs Genmo Mochi 1 test →A dolphin leaps gracefully above crystal-clear waves under a vibrant sunset, the orange and pink hues shimmering on the water's surface as seabirds glide overhead.
Preferred over Pika 2.2 by 95% of raters on this shared prompt.
See the full Google Veo 3.1 vs Pika 2.2 test →A bustling kitchen where chefs expertly prepare gourmet dishes. The camera zooms in on sizzling pans and colorful ingredients, capturing the precision and artistry in each step as flames dance and flavors mingle.
Preferred over Seedance 1 Pro by 95% of raters on this shared prompt.
See the full Google Veo 3.1 vs Seedance 1 Pro test →A cinematic close-up of a barista crafting latte art in a bustling coffee shop. The scene alternates between her focused, skilled hands and customers watching appreciatively, highlighting the artistry and dedication in everyday routines.
Preferred over Wan 2.1 by 95% of raters on this shared prompt.
See the full Google Veo 3.1 vs Wan 2.1 test →Clips and preference votes from Rapidata's text-to-video human preference datasets. Preference values are pairwise shares for that exact comparison, not absolute quality scores.
要点： Sora was discontinued in 2026 — the app closed on April 26 and the API on September 24 — so Veo 3.1 is the available side of this pairing: published per-second API pricing, native audio, 4K and scene extension. The shared-prompt clips and preference split remain as a record of how the two compared.
比較の詳細を読む →Shared-prompt test
69 shared prompts with the human preference split
Shared-prompt test
69 shared prompts with the human preference split
Shared-prompt test
69 shared prompts with the human preference split
Shared-prompt test
69 shared prompts with the human preference split
Shared-prompt test
68 shared prompts with the human preference split
Shared-prompt test
68 shared prompts with the human preference split
Shared-prompt test
67 shared prompts with the human preference split
Shared-prompt test
67 shared prompts with the human preference split
Shared-prompt test
66 shared prompts with the human preference split
Shared-prompt test
66 shared prompts with the human preference split
Shared-prompt test
64 shared prompts with the human preference split
Shared-prompt test
62 shared prompts with the human preference split
See all head-to-heads in the arena hub.
It is the newest Veo. Google's docs now position Gemini Omni Flash as the default video model at roughly $0.10 per effective 720p second, while Veo 3.1 remains the option for native audio, 4K, ingredients, first–last frame control and scene extension.
Per output second with audio included: Standard $0.40 at 720p and 1080p ($0.60 at 4K), Fast $0.10–$0.30 by resolution, Lite $0.05–$0.08 with no 4K. There is no free tier; Google states you are billed only when a video actually generates.
Single clips run 4, 6 or 8 seconds, and 1080p or 4K output requires the 8-second setting. Scene extension adds 7 seconds per step for up to 20 steps — about 148 seconds total from an input of 141 seconds or less.
They share the same 3.1 capabilities at different price and quality points. Lite is the budget tier but drops 4K, ingredients-to-video and scene extension; Standard is the full-feature tier; Fast trades some fidelity for speed between the two.
Yes — audio is natively generated with every clip, cannot be turned off, and is included in the per-second price.
The Arena matchups linked below run public shared-prompt tests: each pairing shows both models' clips for the same prompt with a human preference split published by Rapidata. This site hosts none of that media and adds no rating of its own.
No. Veo has no free tier on the Gemini API, and Google states you are billed only when a video actually generates. The cheapest checked rate is Veo 3.1 Lite at $0.05 per output second, so an 8-second 720p clip costs $0.40 at list price. Consumer AI plans include Flow credits, but Google's plans page lists Omni Flash as the video model those credits run.
Gemini API Veo docs
対象範囲： Official Veo 3.1 reference: model IDs, 4/6/8s durations, resolutions, ingredients, first–last frame, extension (checked 2026-09-28).
Google Gemini API pricing
対象範囲： Official per-second tier pricing, audio included, no free tier (checked 2026-09-28, unchanged).
Google DeepMind Veo model page
対象範囲： Official DeepMind Veo overview.
Google Cloud: ultimate prompting guide for Veo 3.1
対象範囲： Official prompt structure guidance: camera, subject, action, scene, style, audio cues and dialogue in quotes.
確認日 2026-09-28