2026-11-09 · VideoFreeTier

Veo 3.1 Lite free tier: 4s and 8s cost the same 10 credits, and its Gemini API model ID was shut down on 22 October 2026

Veo 3.1 Lite is billed by the clip, not by the second. Read on 9 November 2026, the Google Flow help centre puts Generate 4s, 6s, or 8s videos on a single row at 10 credits for All other users and 5 credits for Google AI Ultra subscribers, with a third row at 0 credits in Lower Priority mode; the header above them reads Credit cost per generation and there is no seconds column anywhere in the table. In a separate channel, veo-3.1-lite-generate-preview in the Gemini API was shut down on October 22, 2026, while the paid Gemini Enterprise Agent Platform still lists Veo 3.1 Lite for sale. Every clip count and every per-second figure on this page is our arithmetic from published numbers, not a Google-published price.

Everything below was read on 9 November 2026 from public Google pages in their en rendering (?hl=en) without registering, subscribing, buying credits, or running a generation. Three different channels are in play here and none of them speaks for the others: the Google Flow consumer plan, the paid developer platform, and the Gemini API. Where a figure belongs to a channel rather than to a model, this page says so. Where an official page prints nothing, this page says so instead of filling the gap. The dated observations behind the model pages live on the data page.

The row that makes it per clip

Manage your Google Flow credits (support.google.com/flow/answer/16526234?hl=en) opens its Lite section with With Veo 3.1 Lite, all users can generate 4s, 6s, or 8s videos in 720p. and Credit costs per generation depend on your Google AI subscription plan. The table that follows has four columns, Feature, Eligible users, Resolution and Credit cost per generation, and three rows, quoted as printed on 9 November 2026:

One row, three lengths, one price. The table offers no per-second rate for Lite and prints no different number for 4 seconds against 8 seconds. The same help page states With or without a subscription, you receive 50 Google Flow credits per day. and, immediately before its retirement table, The Google Flow/AI credit costs below are per generation, not per request. followed by Some product features will create multiple generations per request. A single button press can therefore spend more than one generation, and the unit the page gives is the generation.

Google's own model help page supplies the contrast in plain words. Under Learn about supported features in Omni Flash 1.2 it says Use Omni Flash 1.2 for higher visual quality and consistency, longer video lengths, and duration-based pricing. Its Omni table then prints 1 credit per second at 360p and 3 credits per second at 720p. So two models on the same consumer plan are metered in two different units: Omni Flash 1.2 by duration, Veo 3.1 Lite by the clip. That is a channel fact about how Google Flow charges, not a statement about what either model's weights can do. An earlier guide on this site, Google Flow free credits, worked through the per-second side of that table; this page takes the per-clip side.

What 50 credits buy when length stops mattering (our arithmetic)

With the daily 50 credits and a flat 10 credits per generation for All other users, the number of clips does not change when the length changes. Every cell below is our arithmetic: 50 divided by 10, then multiplied by the chosen length. No Google page publishes any of these clip counts or totals.

So the same 50 credits buy five clips whichever length you pick, and choosing 8 seconds every time gets 40 seconds of video instead of 20 at no extra cost. For contrast, the same 50 credits spent on Omni Flash 1.2 at 720p (3 credits per second) cover about 16 seconds, our arithmetic again, and a different model on the same plan. The five-clips-a-day figure is not new on this site: Veo 3.1 free tier opened with it back in September while making a different argument about version numbers. What is new here is the unit behind it: the reason five does not become ten when you halve the length is that length was never in the Lite price.

The same Lite row appears twice on one help page

The points page does something easy to miss. It has a standalone section headed Veo 3.1 Lite credit costs with the three rows above, and then a separate table headed Retiring Google Flow model credit costs that also lists Lite: Veo 3.1 - Lite, 4s, 6s, 8s videos, Non-Ultra subscribers: 10 and Ultra subscribers: 5. The two places agree on the numbers, 10 and 5, but the second table's heading says retiring while the page's own retirement notice names three other models and points users towards Lite.

Here is that notice as printed: On December 5, 2026, Veo 3.1 Fast, Veo 3.1 Quality, and Omni Flash 1.1 will be retired. Until then, you can generate with these models at standard credit rates while transitioning your workflows to Omni Flash 1.2 or Veo 3.1 Lite. The model help page says the same in slightly different words: we recommend switching to Omni Flash 1.2 or Veo 3.1 Lite for future projects. Both notices also record On November 5, 2026, Veo Extend was retired. So Lite is named as a destination in the retirement text and as a row inside the table titled retiring. This page keeps both statements side by side and does not reconcile them; no official sentence on either page explains why the Lite row sits under that heading. Both were read on 9 November 2026, before the 5 December date.

Zero credits, and the one line Google does not explain

The third Lite row is the strangest. Generate 4s, 6s, or 8s videos in Lower Priority mode is priced at 0 credits and restricted to Ultra 20x subscribers & family plan managers only. Google publishes no separate explanation of what Lower Priority mode does. Across the Flow help articles captured on 9 November 2026, the phrase lower priority appears exactly once, in that row, and no help page describes a queue, a speed, or a wait for it. This page therefore does not claim that zero-credit generations are slower, throttled, or inserted into any particular queue, because none of those mechanisms is printed.

There is one nearby sentence that stops short of explaining it. A different help article, answering Why do I see "You're requesting generations too quickly"?, says we may rate-limit the number of generations you can make, particularly for zero-credit models. That sentence does not name Lower Priority mode, does not name Veo 3.1 Lite, and does not connect itself to the 0-credit row, so we quote it as an adjacent statement and not as a definition. The help pages also use two names for the eligible plan, Ultra 20x subscribers & family plan managers only on the points page and Google One AI Ultra 20x Plan in that other article, but none of these pages prints a subscription price beside either name. That correspondence is 未公布 / 本站未实测 here.

On the paid developer platform the unit is 1 count

The Gemini Enterprise Agent Platform pricing page (cloud.google.com/gemini-enterprise-agent-platform/generative-ai/pricing?hl=en, read in its en rendering on 9 November 2026) has a Veo table whose columns are Model, Feature, Description, Input, Output, Output Resolution and Price (USD). There is no duration column. Its Veo 3.1 Lite rows, quoted as printed:

The page also states Prices are listed in US Dollars (USD). On the same table, Veo 3.1 is $0.40 / 1 count at 720p and 1080p for Video + Audio, and Veo 3.1 Fast is $0.10 / 1 count at 720p. Two boundaries are worth keeping straight. This platform offers Lite at 1080p, while the Flow consumer page says in 720p only; that is a difference between channels, not a contradiction inside one price list. And 1 count is not a second: the table has no length column and no official sentence on the captured page defines what one count contains. A third-party video title reads Veo 3.1 Lite on GMI Cloud — $0.05 per Request (the rest of that title says New Model), printing the same number with a per-request unit; this site did not create an account there, did not measure its billing, and draws no conclusion about whether that is a resale, a discount, or a coincidence.

Any per-second number from this table is ours and depends on an assumption. If, and only if, one count were an 8-second clip, the longest length Lite offers in Flow, then $0.05 ÷ 8 = $0.00625/s (our arithmetic). The Gemini API pricing page prints an effective price of approximately $0.07 per second for gemini-omni-1.2-flash at 720p, which would put the two about 11 times apart (our arithmetic). Both are our conversions across two different paid products, and the second is itself described by Google as an effective price derived from token consumption, not a posted per-second rate. Writing $0.05 / 1 count as $0.05 per second would be wrong by a factor of eight under that assumption; we do not do it.

The Gemini API model ID is gone; the paid platform still sells Lite

The Gemini API video overview (ai.google.dev/gemini-api/docs/video?hl=en) opens with The Veo 3.1 Preview in the Gemini API has been shut down. and, under Veo 3.1, prints the full sentence: Veo 3.1 Preview (veo-3.1-generate-preview, veo-3.1-fast-generate-preview, and veo-3.1-lite-generate-preview) in the Gemini API was shut down on October 22, 2026; migrate to Gemini Omni Flash (gemini-omni-1.2-flash) or the GA models available through the Gemini Enterprise Agent Platform to avoid service interruptions. The date and all three model IDs are quoted as printed on 9 November 2026.

What was shut down is the Preview endpoint inside the Gemini API, including the Lite model ID, not the Veo 3.1 Lite name everywhere. On the same day, the Gemini API's own pricing page and models page, rendered in full, contain zero occurrences of veo; their video entries are gemini-omni-1.2-flash and gemini-omni-1.1-flash. Meanwhile the enterprise platform pricing table described above still carries a live Veo 3.1 Lite section with prices. So veo 3.1 lite api, api pricing and api key are searching for an endpoint that one Google developer surface has retired, while a different Google paid surface still lists the model. That is a channel-level distinction: nothing here says the model itself was discontinued, and nothing here says the consumer Flow plan changed. An earlier guide, Is Gemini Omni Flash free, covered which Google surfaces name Omni at all; this page does not revisit those three surfaces.

The capability table says Lite does have ingredients

Learn about Google Flow models & supported features (support.google.com/flow/answer/16352836?hl=en) introduces Lite as Use Veo 3.1 Lite as a dependable, low-cost baseline for standard 720p creation. Its supported list, as printed on 9 November 2026:

Under Unsupported features in Veo 3.1 - Lite the page lists exactly two items: Video to Video editing and Extend videos. Ingredients and references are not on that list; they are supported, but at 8 seconds only. Notably, the Veo 3.1 Fast section one heading below prints the identical supported and unsupported lists, while the Veo 3.1 Quality section is the one where Ingredients/References to Video appears under unsupported features.

Comments under comparison videos are quoted here as reader statements, not as product documentation, and several of them describe a narrower model. One reader wrote Unfortunately Lite doesn't allow Ingredients to Video, which is my prefered way to work.; another In image-to-video Lite looks better than Fast, but it does not have ingredients-to-video yet.; a third You can't use reference function with Lite model. This is a serious downgrade. Against them, another reader described the opposite workflow: I just made an AI Music Video on my channel using just Veo 3.1 Lite ... give it a Character Sheet + The environment as ingredients. and a fifth said The only thing i get out of it is first frame last frame. We did not run any of these workflows. The official table is the only thing on this page that states what is supported; the comments show that the impression of Lite as a stripped-down model is still circulating, and we do not date or explain when it arose.

The same comment threads use point numbers that do not match the help centre table. One reader wrote 20 creds for veo 2...sorry 3.1 lite ... and 3.1 fast = 40 credits, another referred to The 80-credit price difference simply isn't worth it right now., and a third simply asked What is the cost difference between Lite and Quality? Thanks. The points page prints Lite at 10 for Non-Ultra subscribers and 5 for Ultra subscribers in its retirement table, and Veo 3.1 - Quality at All users: 100. Against those rows, a Lite-to-Quality gap would be 90 credits for a non-Ultra user (our arithmetic), not 80, and the 20 and 40 figures are twice the printed Lite and Fast non-Ultra numbers. This page puts the two sets side by side and does not explain the difference: no official page in this capture says the commenters were using an older table, a different plan, or a different region.

Audio: the paid platform says it, the consumer help pages do not

The enterprise pricing table puts Lite under Video + Audio generation and describes that row as Generate videos with synchronized speech/sound effects from a text prompt or reference image the fastest. That is an explicit audio claim from the paid developer platform. On the consumer side, of the Flow help articles captured on 9 November 2026, exactly two name Veo 3.1 Lite, the points page and the model help page, and neither gives Lite an audio line in its cost table or its supported-features list. The model help page's Create custom voices: item sits inside a paragraph headed All users can access Omni Flash models, which also offer these exclusive features:, not inside the Lite section.

That is as far as the printed pages go. Other Flow help articles do mention audio in other contexts, an Audio Generation Failed troubleshooting question, Flow Music, avatar likeness, and sound on hover over an asset, and one says Sometimes Veo can produce low-quality audio without naming a Lite variant. None of those sentences attributes audio to Veo 3.1 Lite. A reader in one comment thread wrote and no voice, but this page does not treat a comment as a measurement: whether a Lite generation in the consumer Flow app actually contains sound is 未公布 / 本站未实测. The platform-side label and the consumer-side silence are kept side by side.

What this page does not settle

No account was created, no subscription bought, no credit balance observed, no API key used, and no generation run. So this page does not measure whether one count on the enterprise platform changes with clip length, what Lower Priority mode actually does to a queue, whether a Lite clip in Flow carries audio, what plan the Ultra 20x label maps to in dollars, or whether a single request that creates multiple generations is charged once or several times in practice. Those are 未公布 / 本站未实测 where the pages are silent, and nothing here states a view on watermarks or commercial licensing.

Every number above belongs to a named channel: the 5, 10 and 0 credit rows and the 50 daily credits to the Google Flow plan and its help centre, the 1 count prices to the Gemini Enterprise Agent Platform, and the shutdown to the Gemini API's Preview endpoint. None of them is a property of the model weights, and a statement about one channel is not a statement about the others. All official pages were read in their en rendering (?hl=en) on 9 November 2026; pricing and feature wording can vary by plan, platform and region, so a reader in another market should not expect the exact same text. Suggestion lists and comment threads only tell us which phrasings people use; this page quotes no ranking, volume, difficulty, bid, or traffic figures. The dated observations behind the model pages live on the data page.

Related on this site: Google Flow free credits — the per-second table for Omni, where Lite is one row. Veo 3.1 free tier — what a version number does to a free-tier answer. Is Gemini Omni Flash free — which Google surfaces name the model at all. The observed figures behind the model pages live in the data page.