# Cheapest Video Generation API: What a 10-Second Clip Costs

> As of November 12, 2026, a 10-second 720p clip costs $0.30 flat on Gemini Omni 1.2 Flash or Muse via LaoZhang API, and up to $1.20 on other video APIs.

- URL: https://blog.laozhang.ai/en/posts/cheapest-video-generation-api
- Published: 2026-11-12
- Updated: 2026-11-12
- Author: LaoZhang AI Team (https://blog.laozhang.ai/en/about)
- Topic: AI Video Generation
- Tags: Video Generation API, AI Video API Pricing, Gemini Omni 1.2 Flash, Muse Video, MiniMax H3

---
As of November 12, 2026, xAI lists `grok-imagine-video-1.5-lite` at $0.02 per second, which would make a 10-second clip $0.20, the smallest figure in the table below. xAI prints that one rate for the model, but its [video generation guide](https://docs.x.ai/docs/guides/video-generation) says both duration and resolution affect the total, and the model defaults to 480p, so check the charge on a 720p job before you budget around it. Among the other APIs priced below, the lowest fixed price for a 10-second 720p clip is $0.30, on Gemini Omni 1.2 Flash or Meta's Muse through LaoZhang API. For Omni, that is 43% of the roughly $0.70 Google lists for the same 10-second clip.

*Disclosure: [laozhang.ai is our own service](https://blog.laozhang.ai/en/about#laozhang). Its prices are platform pricing, not an independent comparison.*

So for clips of 8 to 10 seconds, send `gemini-omni-1.2-flash` (characters speak your lines, and 1080p costs $0.45) or `muse-video` (the same $0.30 for 5, 8 or 10 seconds, with single clips up to 60 seconds on the longer models) to `https://api.laozhang.ai/v1/videos`. For clips of about 5 seconds, a per-second price wins instead: MiniMax H3 through LaoZhang API costs $0.05 a second at 768P, or $0.25 for 5 seconds, which is 37.5% below MiniMax's own $0.08 list price.

Every figure here is a list price for one text-to-video generation with no reference video, taken from each vendor's pricing page and recomputed for 5, 8, 10 and 15 seconds on November 12, 2026. A shot that takes three attempts costs three times as much, and prices change often, so recheck the vendor page before you commit.

## What one 10-second 720p clip costs on each video API

Ranked by the cheapest way to get a 10-second clip at 720p, or 768P where MiniMax has no 720p tier, the spread runs from $0.20 to $1.20, a factor of six. Per-second prices are multiplied by 10; per-clip prices are the vendor's own figure for that length. The LaoZhang API column covers Gemini Omni 1.2 Flash, Muse and MiniMax H3.

*Disclosure: [laozhang.ai is our own service](https://blog.laozhang.ai/en/about#laozhang). Its prices are platform pricing, not an independent comparison.*

| Model (vendor), resolution | How it's billed | 10 s at the vendor's list price | 10 s through LaoZhang API |
| --- | --- | --- | --- |
| grok-imagine-video-1.5-lite (xAI), 720p available, 480p default | $0.020 per second, no separate 720p rate | $0.20 | — |
| Gemini Omni 1.2 Flash (Google), 720p | Google: 5,792 tokens per second at $12 per 1M, about $0.07 a second. LaoZhang API: per clip, same price from 3 to 10 s | About $0.70 | $0.30 |
| Muse (Meta), 1280×720 | Per clip, 5, 8 or 10 s | Meta has no video API | $0.30 |
| Hailuo-2.3-Fast (MiniMax, legacy), 768P | Per video | $0.32 | — |
| MiniMax-H3-Lite, 768P | $0.04 per second | $0.40 | — |
| MiniMax-H3, 768P | MiniMax: $0.08 per second. LaoZhang API: $0.05 per second | $0.80 | $0.50 |
| grok-imagine-video (xAI), up to 720p | $0.050 per second | $0.50 | — |
| gen4_turbo (Runway) | 5 credits per second at $0.01 | $0.50 | — |
| Hailuo-2.3 (MiniMax, legacy), 768P | Per video | $0.56 | — |
| MiniMax-H3-Max, 768P | $0.08 per second | $0.80 | — |
| grok-imagine-video-1.5 (xAI) | $0.080 per second | $0.80 | — |
| Ray3.2 (Luma), 720p | 300 credits per 10 s clip | About $0.90 | — |
| wan3.0-video (Alibaba Cloud, international), 720P | $0.10 per second list price | $1.00 | — |
| Gen-4.5 (Runway) | 12 credits per second at $0.01 | $1.20 | — |

Official prices come from the [Gemini API pricing page](https://ai.google.dev/gemini-api/docs/pricing), [MiniMax pay-as-you-go pricing](https://platform.minimax.io/docs/guides/pricing-paygo), the [xAI models page](https://docs.x.ai/docs/models), [Runway's API pricing](https://docs.dev.runwayml.com/guides/pricing/), [Luma's API pricing](https://lumalabs.ai/api/pricing) and [Alibaba Cloud Model Studio pricing](https://www.alibabacloud.com/help/en/model-studio/model-pricing), all as of November 12, 2026. A few rows need a caveat. Luma lists credits only; its FAQ gives $0.30 for the 100-credit 5-second 720p job, which puts a credit at $0.003 and the 10-second clip at about $0.90. Runway's pricing page doesn't state the output resolution or allowed lengths for gen4_turbo and Gen-4.5. MiniMax doesn't print a maximum length for H3-Lite on its pricing page, and Alibaba Cloud shows a "limited-time 30% off" tag on Wan 3.0 without printing the discounted rate.

That leaves $0.30 for a 10-second 720p clip on Gemini Omni 1.2 Flash or Muse as the lowest flat per-clip price in this table. It is also below every other 720p list price in the table except xAI's lite model, whose single listed rate isn't broken out by resolution. Veo 3.1 can't make a 10-second clip in one request, because it generates 4, 6 or 8 seconds at a time; its rates are in [Veo 3 Price 2026: Cloud API Pricing, Flow Credits, Cost per Video](https://blog.laozhang.ai/en/posts/veo-3-price). Seedance bills by tokens, and its per-resolution rates are in [Seedance 2.5 API Pricing: Cost per Second at 480p, 720p, 1080p](https://blog.laozhang.ai/en/posts/seedance-2-pricing-free-vs-paid-guide).

## Per clip or per second: which is cheaper at 5, 8, 10 and 15 seconds

A flat per-clip price gets cheaper per second as the clip gets longer, while a per-second price grows in a straight line, so the cheaper option depends on your clip length. The break-even length is the clip price divided by the per-second rate. A $0.30 clip beats $0.05 per second from 6 seconds up, beats $0.04 per second from 7.5 seconds, and never beats xAI's listed $0.02 within that model's 15-second limit.

![Line chart of clip cost by length: the flat $0.30 clip crosses $0.05 per second at 6 seconds and $0.04 per second at 7.5 seconds, while $0.02 per second only reaches $0.30 at 15 seconds](https://blog.laozhang.ai/posts/en/cheapest-video-generation-api/img/per-clip-vs-per-second-break-even.webp)

| Model and price source | 5 s | 8 s | 10 s | 15 s |
| --- | --- | --- | --- | --- |
| grok-imagine-video-1.5-lite, xAI list rate | $0.10 | $0.16 | $0.20 | $0.30 |
| MiniMax-H3-Lite 768P, MiniMax list | $0.20 | $0.32 | $0.40 | Max length not listed |
| MiniMax H3 768P, LaoZhang API | $0.25 | $0.40 | $0.50 | $0.75 |
| grok-imagine-video, xAI list | $0.25 | $0.40 | $0.50 | $0.75 |
| Gemini Omni 1.2 Flash 720p, LaoZhang API | $0.30 | $0.30 | $0.30 | Not available, 10 s max |
| Muse 720p, LaoZhang API | $0.30 | $0.30 | $0.30 | Not available; 20 s costs $0.55 |
| MiniMax-H3 768P, MiniMax list | $0.40 | $0.64 | $0.80 | $1.20 |

*Disclosure: [laozhang.ai is our own service](https://blog.laozhang.ai/en/about#laozhang). Its prices are platform pricing, not an independent comparison.*

Read the table by your own clip length. At 5 seconds, MiniMax H3 through LaoZhang API ($0.25) undercuts the $0.30 flat clip. At 8 seconds the flat clip is already ahead of H3's $0.40, and at 10 seconds it costs $0.03 a second, less than any per-second rate except xAI's lite model. Between 11 and 15 seconds the flat options drop out, because Gemini Omni 1.2 Flash through LaoZhang API and `muse-video` both stop at 10 seconds; MiniMax H3 runs to 15 seconds ($0.75 at 768P). MiniMax itself still prices its legacy Hailuo 2.3 models per video: Hailuo-2.3-Fast is $0.19 for 6 seconds and $0.32 for 10 seconds at 768P.

Past 15 seconds, Muse is the only model here that returns one continuous clip from a single request. Through LaoZhang API, `muse-video-20s` costs $0.55, `muse-video-30s` $0.80 and `muse-video-60s` $1.55, which works out to $0.0275 a second at 20 seconds and about $0.026 at 60. How Gemini Omni's flat clip compares with Google's per-second rate at shorter lengths is worked out in [Gemini Omni 1.2 Flash API Gateway: 4 Routes, Cost by Clip Length](https://blog.laozhang.ai/en/posts/gemini-omni-1-2-flash-api-gateway).

These tables were recomputed on November 12, 2026 with a short Python script from the list prices above (clip cost = per-second rate × seconds; break-even = clip price ÷ per-second rate). They count one generation per clip. Monthly plans, retries and the cost of a clip you actually keep are covered in [How Much Does an AI Video Generator Cost Per Month and Per Clip?](https://blog.laozhang.ai/en/posts/how-much-does-ai-video-generator-cost).

## Which cheap video API fits your clip

Price only settles the choice once the clip meets your requirements, and the cheapest options differ in length, resolution and sound. If characters need to speak lines from your prompt, use Gemini Omni 1.2 Flash: through LaoZhang API it costs $0.30 at 720p and $0.45 at 1080p for any length from 3 to 10 seconds, and every clip has its own audio. That route stops at 10 seconds, 1080p and image references; Google's own Gemini API goes to 15 seconds and upscaled 4K at its per-second price, as listed on the [Gemini Omni documentation](https://ai.google.dev/gemini-api/docs/omni).

*Disclosure: [laozhang.ai is our own service](https://blog.laozhang.ai/en/about#laozhang). Its prices are platform pricing, not an independent comparison.*

If you need one continuous shot longer than 15 seconds, Muse is the option, at $0.55 for 20 seconds up to $1.55 for 60, always at the 720p tier with music and sound effects. If you need clips of 5 seconds or less, 2K output, or a reference video that steers camera movement, MiniMax H3 fits better: through LaoZhang API it costs $0.05 a second at 768P and $0.08 at 2K (MiniMax lists $0.13 at 2K), runs 4 to 15 seconds, and bills a reference video's seconds at the same rate. It uses the native `/hailuo/v2` endpoints, not `/v1/videos`; per-clip costs at both resolutions are worked out in [MiniMax H3 (Hailuo) API Pricing: Cost Per Clip at 768P and 2K](https://blog.laozhang.ai/en/posts/minimax-h3-api-pricing).

If a draft resolution is good enough, the price drops further. Gemini Omni 1.2 Flash at 360p costs $0.10 per clip through LaoZhang API, MiniMax lists Hailuo-02 at $0.15 for a 10-second 512P video, and MiniMax-H3-Lite at 480P is $0.025 a second. Test prompts at a low resolution and render only the keepers at 720p or 1080p.

![Which cheap video API fits each clip: Gemini Omni 1.2 Flash at $0.30 for spoken lines, Muse at $0.55 for 20 seconds, MiniMax H3 at $0.05 per second for short or 2K clips, and Omni at 360p for $0.10 drafts](https://blog.laozhang.ai/posts/en/cheapest-video-generation-api/img/which-video-api-fits-your-clip.webp)

### Image to video: does a start image change the price?

Mostly not. Through LaoZhang API, Gemini Omni 1.2 Flash costs the same whether you send a text prompt, a first-frame image or up to six images in total, and Muse's price doesn't change when you add a first frame or reference images. MiniMax H3 bills the same per-second rate with a first frame, and its first five reference images are free, then $0.03 each. Among the official price lists, Luma prices text-to-video and image-to-video on the same row. For a comparison of image-to-video models by output rather than price, see [AI Video Generator from Image: Best Image-to-Video Routes and When to Switch](https://blog.laozhang.ai/en/posts/best-ai-image-to-video-generator).

## Wait times and same-prompt clips through LaoZhang API

A cheap clip that takes minutes to finish changes how you design a queue, so here are the times measured in LaoZhang API test runs. All three models are asynchronous: you create a task, poll it, then download the MP4. Each figure comes from one test run, timed from task creation to completion, and is not an average or a promise.

*Disclosure: [laozhang.ai is our own service](https://blog.laozhang.ai/en/about#laozhang). Its prices are platform pricing, not an independent comparison.*

| Model | Clip | Price | Finished in | Test date |
| --- | --- | --- | --- | --- |
| Gemini Omni 1.2 Flash | 720p, 10 s | $0.30 | 67.2 s (about 53 s on November 11) | November 12, 2026 |
| Muse (`muse-video`) | 1280×720, 10 s | $0.30 | 154 s | November 2, 2026 |
| MiniMax H3 | Five clips, 768P and 2K, 6–10 s | $0.05–$0.08 per second | 50–103 s each | November 2, 2026 |

Charges in these runs matched the list prices. To compare output for the money, the same prompt, "A paper boat drifts down a rainy city gutter, then sails into a small sunny pond", was run at 16:9 on all three models: [Gemini Omni 1.2 Flash at 720p, 10 seconds](https://media.laozhang.ai/samples/video/gemini-omni-paper-boat-720p-10s.mp4) for $0.30, [MiniMax H3 at 768P, 10 seconds](https://media.laozhang.ai/samples/video/h3-paper-boat-768p-10s.mp4) for $0.50, and [Muse at 20 seconds with `muse-video-20s`](https://media.laozhang.ai/samples/video/muse-paper-boat-20s.mp4) for $0.55. The clips are unedited, with the sound each model generated; there is no 10-second Muse clip of this prompt.

## Can I get guaranteed concurrency from an AI video API?

Not as a guarantee. As of November 12, 2026, LaoZhang API's [concurrency FAQ](https://docs.laozhang.ai/en/faq/api-concurrency) says it doesn't limit concurrency or throttle requests per account or token for video models, and that any change will be posted in its announcements. Under heavy load a request can still come back with a `429`, which isn't billed, so cap in-flight jobs with a local queue and back off before you retry.

*Disclosure: [laozhang.ai is our own service](https://blog.laozhang.ai/en/about#laozhang). Its prices are platform pricing, not an independent comparison.*

In a test run on November 11, 2026, with 12 requests in flight at once, 13 out of 13 Gemini Omni 1.2 Flash tasks completed through LaoZhang API, each in 30 to 58 seconds, and every charge matched the list price. There are no measured figures yet for more than 12 jobs at once, or for parallel Muse and MiniMax H3 jobs. LaoZhang API's docs don't publish an SLA for throughput or latency; the [status page](https://api.laozhang.ai/uptimeStatus) shows live availability of each endpoint.

For video jobs, the expensive mistake is resubmitting a task that is still running, because every new task is charged again. Poll the status instead: every 5–10 seconds for Gemini Omni, every 15 seconds for Muse and every 10 seconds for MiniMax H3, as their LaoZhang API guides recommend.

## Make a 10-second clip for $0.30 with one request

Gemini Omni 1.2 Flash and Muse share the same `POST /v1/videos` endpoint on LaoZhang API, so one request covers both; change `model` to `muse-video` for Muse. Set your API key first:

*Disclosure: [laozhang.ai is our own service](https://blog.laozhang.ai/en/about#laozhang). Its prices are platform pricing, not an independent comparison.*

```bash
export LAOZHANG_API_KEY="your LaoZhang API key"

curl -s -X POST "https://api.laozhang.ai/v1/videos" \
  -H "Authorization: Bearer $LAOZHANG_API_KEY" \
  -H "Content-Type: application/json" \
  --data-raw '{
    "model": "gemini-omni-1.2-flash",
    "prompt": "A paper boat drifts down a rainy city gutter, then sails into a small sunny pond",
    "resolution": "720p",
    "seconds": "10",
    "aspect_ratio": "16:9"
  }'
```

The response returns a task `id` with `status: "queued"`. Poll `GET https://api.laozhang.ai/v1/videos/{id}` until `status` is `completed`, then download the MP4 from `video_url`; the link stays available for 7 days. Full request fields, error messages and the Python version are in LaoZhang API's [Gemini Omni 1.2 Flash video API guide](https://docs.laozhang.ai/en/api-capabilities/gemini-omni-video-generation) and [Meta Muse video API guide](https://docs.laozhang.ai/en/api-capabilities/muse-video-generation).

## Cheap video generation API: other questions

### Is there a free video generation API?

Not as an ongoing free tier from the major vendors. As of November 12, 2026, Google's Gemini API lists no free tier for Gemini Omni 1.2 Flash, and Alibaba Cloud's international Model Studio lists 30 seconds of free quota for wan3.0-video ahead of its $0.10-a-second 720P rate. Free trials and free apps that exist are compared in [Is There a Free AI Video API? What Works Without Paying](https://blog.laozhang.ai/en/posts/best-free-ai-video-api).

### Do I pay for failed or blocked videos?

Not through LaoZhang API. A failed video task is refunded in full (for Gemini Omni 1.2 Flash this includes tasks blocked by content moderation), and a request rejected at submission with a parameter error isn't charged, according to LaoZhang API's Gemini Omni, Muse and MiniMax H3 guides as of November 12, 2026. A clip that renders but that you don't use is still charged.

*Disclosure: [laozhang.ai is our own service](https://blog.laozhang.ai/en/about#laozhang). Its prices are platform pricing, not an independent comparison.*

### Which cheap video model gives the best quality?

Price and quality rank differently, and the right model depends on the shot. Model-by-model quality and use cases are compared in [Best AI Video Model in 2026: Complete Comparison Guide (Sora 2, Veo 3.1, Runway & More)](https://blog.laozhang.ai/en/posts/best-ai-video-model).

## Sources

External pages this guide links to, in the order they appear. Last updated 2026-11-12.

- [video generation guide](https://docs.x.ai/docs/guides/video-generation) (docs.x.ai)
- [Gemini API pricing page](https://ai.google.dev/gemini-api/docs/pricing) (ai.google.dev)
- [MiniMax pay-as-you-go pricing](https://platform.minimax.io/docs/guides/pricing-paygo) (platform.minimax.io)
- [xAI models page](https://docs.x.ai/docs/models) (docs.x.ai)
- [Runway's API pricing](https://docs.dev.runwayml.com/guides/pricing/) (docs.dev.runwayml.com)
- [Luma's API pricing](https://lumalabs.ai/api/pricing) (lumalabs.ai)
- [Alibaba Cloud Model Studio pricing](https://www.alibabacloud.com/help/en/model-studio/model-pricing) (alibabacloud.com)
- [Gemini Omni documentation](https://ai.google.dev/gemini-api/docs/omni) (ai.google.dev)