AI video generation API: price comparison (2026)—8 models for the price of one clip

The same 8-second 720p clip costs from $0.18 to $3.03, a difference of about 16.8 times, depending on the model. These are the 8 video models on the same API key. Before the numbers, though, there is one detail that changes the answer: these models do not bill the same way — 3 charge a fixed price per video, while 5 charge per second generated. So “the cheapest API” changes depending on the duration your product generates, and each list's ranking is valid only for the duration it chose without telling you.

Position first: Kunavo is its own product, a gateway that resells access to models, not an independent analysis site. The figures below are taken from the catalog that bills the requests, at the time the page is generated. Kunavo prices verified on October 4, 2026.

What a 8 s 720p clip costs, model by model

The common benchmark is 8 seconds at 720p: this is the unit billed by models priced per video, and all per-second models generate it. The “official price” column shows what the provider itself publishes for the same clip.

A dash in the official price column means we could not confirm a price published by the provider — and we claim no discount for these models. For the 4 models with a published official price, Kunavo is cheaper. We prefer to leave the cell blank rather than fill it with an estimate.

Two billing methods, and why comparisons break

Google Veo 3.1 charges per video, and every Veo video is 8 seconds long, whatever duration is asked for. ByteDance Seedance and Alibaba Wan 2.7 charge per second of output. In practice: at 8 seconds, Veo 3.1 Lite leads at $0.18; ask for 4 seconds and all per-second models cost half as much, while a Veo model still generates and charges for a whole clip. That is why a per-second model that seemed expensive at 8 seconds can be the cheaper of the two at 4. Above 8 seconds, a Veo model needs one more whole clip and a per-second model charges only for the extra seconds, so the per-second models gain again; at 16 seconds each model costs double and the order is the same again. Decide the duration before comparing prices. The AI video cost calculator does the math for your volume.

Which one to choose for each use case

• Drafts, previews, volume — Veo 3.1 Lite, $0.18 per 8 s clip. At this price, discarding a clip costs less than asking someone whether it's worth generating.
• The shot delivered to the client — Veo 3.1 Quality ($1.60 per video), the most expensive fixed-price model, worth it when the result is delivered as is.
• Short clips, 6 seconds or less — per-second models: you pay for what you generate. The lowest per-second price is Seedance 2.0 Mini, at $0.0526 per second.
• Animate existing artwork — Veo 3.1 Lite, Veo 3.1 Fast, Seedance 2.0 Mini, Veo 3.1 Quality, and Seedance 2.5: supports a starting frame, ending frame, and reference image.

What Kunavo does not offer

Other pages in these results list models that are not available on Kunavo: as of September 2026, Sora, Kling, Hailuo, Runway, Luma, and Pika. If your shortlist centers on one of them, this key cannot call it and another provider is the right choice — this page does not try to persuade you otherwise.

How to make a request

The 8 models use the same endpoint; only the model ID changes. Generation is asynchronous: the first request returns a task ID, and the finished video is returned by a later query, because rendering takes tens of seconds — the same structure used by every provider in the sector. The request format is in the video endpoint documentation; Veo details are in Veo 3 pricing. The image side of the same key is in the image generation API, and you can pay with Pix.

Frequently asked questions