Veo 3 API Cost Per Video (2026): Veo 3 Itself Is Gone From the Gemini API — Here Is What an 8, 30 and 60-Second Clip Cos
- Google's Gemini API shut down veo-3.0-generate-001 and the Fast variant on June 30, 2026, pointing both to Veo 3.1. Veo 3 pricing today means Veo 3.1 pricing.
- Veo 3.1 with audio, per Google's Gemini API page on 2026-09-23: Standard $0.40/s (720p/1080p), $0.60 (4k); Fast $0.10-$0.30; Lite $0.05-$0.08. 8 seconds on Standard: $3.20.
- One generation is 4, 6 or 8 seconds. Longer videos come from 7-second extensions at 720p, up to 148 seconds, so 30 or 60 seconds means several billed generations.
- Google's Vertex AI page lists video-only Veo 3.1 at $0.20 instead of $0.40. If you add your own voiceover, turning native audio off is the biggest single saving.
Here is the first thing most Veo 3 pricing articles will not tell you: on the Gemini API, the model called Veo 3 is already gone. Google's Gemini API deprecations page lists veo-3.0-generate-001 and veo-3.0-fast-generate-001 with a shutdown date of June 30, 2026, and names veo-3.1-generate-preview and veo-3.1-fast-generate-preview as the replacements. If you searched "Veo 3 API cost per video" today, the answer you need is the Veo 3.1 answer.
That matters more than it sounds, because the 3.1 family has three tiers, three resolutions and an audio switch, and the same eight seconds can cost anywhere from forty cents to almost five dollars. This page prices real clip lengths using only what Google's own pages showed on 2026-09-23.
What Google actually lists, tier by tier
As listed on the Gemini API pricing page (ai.google.dev), checked 2026-09-23. All prices are per second of video, paid tier, and the page labels them "video with audio price (default)". The free tier column says "Not available" for all three.
| Tier | 720p | 1080p | 4k |
|---|---|---|---|
| Veo 3.1 Standard | $0.40 | $0.40 | $0.60 |
| Veo 3.1 Fast | $0.10 | $0.12 | $0.30 |
| Veo 3.1 Lite | $0.05 | $0.08 | not supported |
Google's Vertex AI generative AI pricing page, checked the same day, lists the same with-audio numbers and adds video-only rows that the Gemini API page does not show: Veo 3.1 at $0.20 (720p/1080p) and $0.40 (4k); Fast at $0.08, $0.10 and $0.25; Lite at $0.03 and $0.05. Vertex prints the unit as "/ 1 count" rather than "per second", while the Gemini API page states per second explicitly; the with-audio figures match exactly. Vertex also still lists Veo 3 ($0.40 with audio, $0.20 video only), Veo 3 Fast and Veo 2 ($0.50), so the two Google surfaces do not retire models on the same schedule. Check which surface your key is on before you read a price.
One line on the Gemini page is worth copying into your own notes: you are only charged if your video is successfully generated. An audio processing failure that blocks the output is not billed.
What an 8, 30 and 60-second clip really costs
This is where "per second" gets slippery. Google's Veo documentation says a single generation is 4, 6 or 8 seconds, and must be 8 when you use 1080p, 4k, reference images or extension. There is no 30-second or 60-second generation. To get there you use video extension: each extension adds 7 seconds, it can run up to 20 times, the combined output tops out at 148 seconds, and extension works on 720p video only. Lite does not support extension at all.
So a 30-second video is one 8-second generation plus four 7-second extensions (36 seconds, trimmed), and a 60-second video is one plus eight (64 seconds). The table assumes each extension is billed on the 7 new seconds at the same per-second rate. The pricing page does not break extension out separately, so treat that as an assumption and confirm it against your first invoice.
| Clip | Seconds generated | Standard 720p | Fast 720p | Lite 720p |
|---|---|---|---|---|
| 4-second, single generation | 4 | $1.60 | $0.40 | $0.20 |
| 8-second, single generation | 8 | $3.20 | $0.80 | $0.40 |
| About 30 seconds (8 + 4 extensions) | 36 | $14.40 | $3.60 | no extension |
| About 60 seconds (8 + 8 extensions) | 64 | $25.60 | $6.40 | no extension |
Two things fall out of this. First, a long Veo video is locked to 720p because extension is, so paying for a 1080p opening clip and then extending at 720p gets you a mixed-resolution file. Second, the jump from Fast to Standard is four times the cost for the whole length. If the Standard look matters for a hero shot, generate that shot on Standard and the rest on Fast, rather than running everything through one tier.
The per-second price is fixed. The number of seconds you pay for is not: extensions, retries and trimmed overhang all add up.
Four switches that cut the bill before you touch the vendor
- Turn native audio off if you replace it anyway. On Vertex's own table, video-only Veo 3.1 is half the with-audio price. If your pipeline adds a voiceover and music afterwards, the generated audio is a line you are paying for and discarding.
- Stop at 1080p for vertical shorts. 4k Standard is $0.60 a second against $0.40. On a phone feed, nobody sees the difference, and the extension feature cannot keep 4k anyway.
- Draft on Lite, finish on Standard. At $0.05 a second, Lite is cheap enough to test prompts and camera direction. Only the prompt that works goes to the expensive tier.
- Count your takes. Google does not charge for failed generations, but it does charge for successful ones you reject. Track how many generations each published clip needed; that multiplier is your real rate.
If you are coming from Sora, keep in mind that Veo is one option among several; the full side-by-side with Kling, Seedance, MiniMax and Runway is in AI video generator API pricing (2026), and the Sora cost story is in Sora 2 leaves the API on September 24. And if the clips are heading to YouTube, the monetization rules care about what you add, not which model rendered it: see the three kinds of video YouTube will not monetize.
Same Veo model, three different bills
There is one more trap in the phrase "Veo 3 API", and it has nothing to do with the version number. The same model family is sold through at least three doors, and each door has its own price table and its own retirement schedule.
- The Gemini API (ai.google.dev). Priced per second, with audio by default, free tier marked "Not available". This is where Veo 3.0 was shut down on June 30, 2026.
- Vertex AI on Google Cloud. Same with-audio numbers for Veo 3.1, plus video-only rows at roughly half the price, and older Veo 3, Veo 3 Fast and Veo 2 rows still on the page. The Gemini API deprecation notice itself points to "the GA models" on Google's enterprise platform as an alternative to the preview models.
- Resellers such as Runway. Runway's API pricing docs list Veo 3.1 at 40 credits per second with audio and 20 credits without, at $0.01 a credit. That is $0.40 and $0.20, the same as Google's Standard rates, so you are paying for convenience and one shared balance, not for a discount.
The practical rule: write down which door your API key opens next to the price you budget with. A team that prototypes on the Gemini API and ships on Vertex can end up with two different model lists and two different invoices for what they think of as one model. And if you only ever need silent footage because a voiceover goes on top, the door that sells video-only generation is the one to use.
Veo 3 API cost: FAQ
How much does one Veo 3 video cost through the API?
Since Veo 3.0 was shut down on the Gemini API on June 30, 2026, the practical answer is Veo 3.1. As listed on Google's Gemini API pricing page on 2026-09-23, an 8-second clip with audio costs $3.20 on Standard (720p or 1080p), $0.80 to $0.96 on Fast, and $0.40 to $0.64 on Lite. 4k is $4.80 on Standard and $2.40 on Fast.
Is there a free tier for the Veo API?
Not on the Gemini API: the pricing page shows "Not available" in the free tier column for all three Veo 3.1 tiers as of 2026-09-23. Free or trial access may exist elsewhere, for example through Google Cloud credits, but that is account-specific and not a published per-second rate.
Veo 3 vs Veo 3 Fast: which should I use?
On current pricing the comparison is Veo 3.1 versus Veo 3.1 Fast, and Fast is a quarter of the Standard price at 720p ($0.10 versus $0.40 per second). A sensible split is Fast or Lite for drafts and most shots, and Standard only for the few shots where the quality difference is visible to your audience. Compare it with other vendors' mid tiers in Kling API pricing before you commit to one family.