Video APIs bill by the second of generated footage, which makes them the most expensive media type to get wrong. Here are the current per-second rates from Google and xAI, and how to estimate a clip's cost before you render it.
Google's Gemini Omni Flash costs $0.10 / s (720p), billed as video output tokens; it replaced the Veo 3.1 Preview models, which shut down on October 22, 2026. xAI sells several Grok Imagine video models priced per second, with a lite tier below its standard models. The real cost is per-second rate × clip length × the number of takes you throw away.
| Model | Provider | Price | Notes |
|---|---|---|---|
| Gemini Omni Flash | $0.10 / s (720p) | GA model (gemini-omni-1.1-flash) replacing the Veo 3.1 Preview line, shut down Oct 22, 2026; billed per output token ($17.50/1M video tok); also offers 360p, 1080p & 4K (upscaled), not separately priced by Google | |
| Grok Imagine 1.5 | SpaceXAI | $0.080 / s | Cheapest frontier tier (480p–1080p) |
| Model | Provider | Price | Notes |
|---|---|---|---|
| Grok Imagine Video | SpaceXAI | $0.050 / s | Older Grok Imagine video model (480p–720p); Grok Imagine 1.5 Lite is now cheaper |
| Grok Imagine 1.5 Lite | SpaceXAI | $0.020 / s | Cheapest Grok Imagine video tier (grok-imagine-video-1.5-lite) |
Multiply the per-second rate by the clip length, then by how many attempts a usable shot takes. A 10-second clip at $0.10 / s (720p) costs ten times the per-second rate; if your prompts need three takes on average, budget thirty seconds of output per finished clip. Draft at the lowest resolution and shortest duration that tells you whether the motion works, and only render the final at full length. xAI caps a single Grok Imagine generation at 15 seconds, so longer sequences mean stitching several paid clips.
Gemini Omni Flash (gemini-omni-1.1-flash) bills video as output tokens, which Google converts to about $0.10 / s (720p) at 720p. It can also produce 360p, 1080p and upscaled 4K, but Google doesn't price those separately, so test what other resolutions actually cost on your own account. Google's pricing page shows no free tier and no Batch or Flex discount for this model, so there's no cheap overnight lane for bulk jobs. If you still call Veo 3.1 Preview model IDs, those requests stopped working on October 22, 2026.
xAI's pricing page lists three per-second video models: grok-imagine-video-1.5, the older grok-imagine-video, and a cheaper grok-imagine-video-1.5-lite. Its docs say cost depends on both duration and resolution, so confirm the rate for the resolution you request. Generation is asynchronous (start a job, poll by request ID, fetch the URL), and image-to-video is supported, which lets you lock the first frame with a cheap still from an image API instead of paying for video retries to get the composition right.
Multiply the per-second price by 60. At Gemini Omni Flash's $0.10 / s (720p), one minute of finished 720p video is 60 billed seconds, and more once you count discarded takes.
Gemini Omni Flash (gemini-omni-1.1-flash) is the GA replacement for the Veo 3.1 Preview line, which shut down on October 22, 2026.
Not for Gemini Omni Flash: Google's pricing page lists no free tier for it. Plan on paying for every generated second, including test renders.
xAI's Grok Imagine API generates clips of up to 15 seconds per request. Longer videos are built by generating several clips and editing them together, each billed separately.
How we choose: prices come from each vendor's official pricing page and are re-checked every week. We take no affiliate fees, so rankings aren't for sale.