Video generation APIs
A video generation API takes a prompt, often with a start frame or reference clip, and returns a few seconds of video, more and more often with sound. Almost all of them bill per second of output, so length, resolution and audio set the price, from about 2 cents to 60 cents a second. Google, Runway, Luma, Kling, MiniMax and Alibaba sell their own models. Runway and Pika also resell other labs' models, and fal, Replicate and Together host many labs behind one key, sometimes the same model at a different price. Generate a few clips before you commit: motion, faces and prompt following differ far more than price does. 11 tools, in the order they appear in the directory. This is not a ranking.
Grouped by job: AI Agent Infrastructure
Together AI
Serverless API for open models such as gpt-oss, GLM and Kimi, OpenAI-compatible and billed per token, with batch, provisioned throughput, dedicated endpoints, fine-tuning and GPU clusters when you outgrow shared capacity. Image generation models too, such as FLUX.2, GPT Image 2 and Nano Banana Pro, priced per image. Video generation models as well, such as Seedance 2.5, HappyHorse and Hailuo 02.
fal
Hosted image generation models from many labs behind one API, including FLUX, Nano Banana, Seedream and Z Image, billed per image or per megapixel from prepaid credits, with no charge for server errors or queue time. Video generation models too, such as Kling, MiniMax H3, Seedance, Veo 3.1 and Wan, billed per second or per video, and serverless GPUs for your own models.
Replicate
Run image generation and other models from one API: public models such as FLUX, Ideogram and Recraft are billed per output image or by the second of hardware time, and you can deploy your own models with the open-source Cog. Video generation models too, such as Wan 2.1, billed per second of output video.
Google Veo
Google's Veo video generation models on Google Cloud (the Gemini Enterprise Agent Platform, formerly Vertex AI): Veo 3.1, Veo 3.1 Fast and Veo 3.1 Lite make clips with or without generated audio, from a prompt or a starting image, up to 4K. Billed per second of video in your Google Cloud account.
Gemini Omni Flash
Google's video generation model on the Gemini API, which replaced Veo there: gemini-omni-1.1-flash makes video with audio from text, images and video, edits clips over several turns and extends them, at 360p to 720p with upscaled 1080p and 4K. Billed per token, about 10 cents a second at 720p. Paid tier only.
Runway API
Runway's developer API for video generation: its own Gen-4.5, Gen-4 Turbo, Aleph 2 for editing video and Act-Two for performance capture, plus Veo 3.1, Kling, Seedance, MiniMax, Wan and Gemini Omni Flash on the same key, with a Model Router that picks one for you. Prepaid credits at a cent each, billed per second.
Luma API
Luma's video generation API for Ray3.2: text and image to video up to 1080p, up to 16 keyframes in one clip, video to video up to 20 seconds, Reframe, and HDR with EXR export. Priced per video by length and resolution, from $0.15 for 5 seconds at 540p.
Kling API
Kling's own API for its video generation models: Kling 4.0, 3.0 and 3.0 Turbo, and older 2.x models, from text, images and reference video, with or without audio, up to 4K on Kling 3.0. Billed per second from prepaid resource packages or account balance.
MiniMax Video (Hailuo)
MiniMax's video generation API: the H3 series (H3, H3-Highspeed, H3-Lite and H3-Max) takes text, images, video and audio as input and outputs 480p to 2K, billed per second, with the older Hailuo 2.3 and 02 models priced per clip. LLM, speech and music models on the same account.
Pika API
Pika's developer platform for video generation: its own Pika 2.5 and H3 Flex models next to 100+ others such as Kling, Veo 3.1, Seedance, Wan and Hailuo, billed per second. Member prices need the $10-a-month API Club, which comes back as $10 of credit each month.
Alibaba Wan
Alibaba's Wan video generation models on Alibaba Cloud Model Studio: Wan 3.0, Wan 2.7 and older text, image and reference to video models, plus HappyHorse and partner models such as Vidu and PixVerse, billed per second of output. Wan 2.1 and 2.2 weights are open under Apache 2.0.
What it is, which models, what a second of video costs.
Straight from each vendor's own pages. Plans move often, so the check date is below.
| Tool | What it is | Models | Price per second | Open weights |
|---|---|---|---|---|
| Together AI | Video models on the same Together key as its LLMs, as async jobs. | Seedance 2.5, 2.0 and 1.0, HappyHorse 1.1 and 1.0, Hailuo 02, MiniMax 01 Director, Vidu Q1, PixVerse v5, Wan 2.7. | Seedance 2.5 $0.115 at 480p, from $0.249 at 720p. Other video models are listed with a single price on the pricing page; check the model page for length and resolution. | Some of the models. Together is not. |
| fal | Hosted video models from many labs behind one API, plus serverless GPUs for your own. | Kling 3, MiniMax H3, Seedance 2.0 and 2.5, Veo 3.1, Gemini Omni, Wan 3.0, LTX 2.3, PixVerse V6 and more. | Kling v3 Pro $0.14, Kling 2.5 Turbo Pro $0.07, MiniMax H3 Max $0.03, H3 Max Turbo $0.015. Seedance 2.0 $0.014 per 1,000 tokens. | Some of the models. fal is not. |
| Replicate | Run public models from one API, or deploy your own with Cog. | Wan 2.1 and many other public and proprietary video models. | Wan 2.1 image to video $0.09 at 480p, $0.25 at 720p. Other models priced on each model page or by the second of GPU time. | Some of the models. Replicate is not. |
| Google Veo | Veo on Google Cloud, now called the Gemini Enterprise Agent Platform, billed in your Google Cloud account. | Veo 3.1, Veo 3.1 Fast, Veo 3.1 Lite, Veo 3 and 3 Fast, Veo 2. | Veo 3.1 with audio $0.40 at 720p or 1080p, $0.60 at 4K; video only $0.20. Fast with audio $0.10 at 720p. Lite with audio $0.05 at 720p, video only $0.03. Veo 3 and Veo 2 $0.50. | No. |
| Gemini Omni Flash | Google's video model on the Gemini API, where it replaced Veo. Generates, edits over several turns, and extends clips. | gemini-omni-1.1-flash. | Billed per token: $17.50 per 1M video output tokens, about $0.10 a second at 720p. 360p about $0.034, 1080p about $0.15, 4K about $0.30. Paid tier only. | No. |
| Runway API | Runway's own models plus other labs' on one key, with a Model Router. Prepaid credits at $0.01 each. | Gen-4.5, Gen-4 Turbo, Aleph 2, Act-Two, plus Veo 3.1, Kling, Seedance, MiniMax H3, Wan, HappyHorse and Gemini Omni Flash. | Gen-4.5 $0.12, Gen-4 Turbo $0.05, Aleph 2 $0.28 (56-credit minimum), Act-Two $0.05. Veo 3.1 $0.40 with audio, $0.20 without. | No. |
| Luma API | Luma's Ray3.2 model, with keyframes, video to video, Reframe and HDR. | Ray3.2. | Per clip, not per second: 5 seconds at 540p $0.15, 720p $0.30, 1080p $1.20; 10 seconds $0.45, $0.90, $3.60. Reframe $0.06 a second at 540p. HDR costs twice SDR. | No. |
| Kling API | Kling's own API, billed from prepaid resource packages or account balance. | Kling 4.0, Kling 3.0 and 3.0 Turbo, Kling 2.6 and 2.5 Turbo. | Kling 4.0 $0.157 at 720p, $0.235 at 1080p. Kling 3.0 without audio $0.084 at 720p, with audio $0.126, 4K $0.42. 2.6 and 2.5 Turbo from $0.042. | No. |
| MiniMax Video (Hailuo) | MiniMax's H3 video models, next to its LLM, speech and music APIs. | MiniMax-H3, H3-Highspeed, H3-Lite, H3-Max, and the older Hailuo 2.3, 2.3 Fast and 02. | H3 $0.08 at 768p, $0.13 at 2K. H3-Highspeed $0.04 at 480p. H3-Lite $0.025 at 480p. Input video billed at the same rate. Hailuo 2.3 $0.28 per 6-second 768p clip. | No. |
| Pika API | Pika's own models plus 100+ from other labs. Member prices need the $10-a-month API Club, returned as $10 of credit. | Pika 2.5, Pika H3 Flex, plus Kling, Veo 3.1, Seedance, Wan 3.0, HappyHorse, Hailuo and more. | Pika 2.5 $0.02 at 480p, $0.04 at 720p, $0.06 at 1080p. H3 Flex from $0.01 at 480p to $0.08 at 4K. Wan 3.0 $0.065 at 720p. Member prices include a 5% platform fee. | No. |
| Alibaba Wan | Wan on Alibaba Cloud Model Studio, with HappyHorse and partner models on the same account. | Wan 3.0 and 3.0 Prime, Wan 2.7, 2.6, 2.5 and 2.2, HappyHorse 1.1 and 1.0, Vidu, PixVerse. | Singapore list prices: Wan 3.0 $0.05 at 480p, $0.10 at 720p, $0.20 at 1080p, 30% off for a limited time. Wan 2.7 $0.10 at 720p, $0.15 at 1080p. HappyHorse 1.1 from $0.07. | Wan 2.1 and 2.2, under Apache 2.0. Wan 3.0 is API only. |
Checked 2026-11-02. Sources: Google Cloud generative AI pricing, Veo 3.1 docs, Gemini API pricing, Gemini Omni docs, OpenAI deprecations, Runway API pricing, Luma API, Kling API pricing, MiniMax pricing, Pika API pricing, Alibaba Cloud Model Studio pricing, Wan 2.2 on GitHub, fal pricing, Replicate pricing, Together AI pricing, Together Seedance 2.5 docs.
Pick it if.
- Google VeoYou're on Google Cloud and want Veo with audio up to 4K, from $0.03 a second on Lite.
- Gemini Omni FlashYou already use the Gemini API and want to generate, then edit the same clip over several turns.
- Runway APIYou want Runway's models and a dozen others on one credit balance, or a router that picks for you.
- Luma APIYou need keyframe control or video to video, and like a fixed price per clip.
- Kling APIYou want Kling from the lab that makes it, including 4K on Kling 3.0.
- MiniMax Video (Hailuo)Price per second matters and 480p to 768p is enough, from $0.025 a second.
- Pika APIYou make lots of short clips and $10 a month for member prices pays off.
- Alibaba WanYou may run Wan yourself later and want the hosted version from the team that trains it.
- falYou want to switch between many labs' video models on one bill.
- ReplicateYou want public models and your own Cog-packaged models on one API.
- Together AIYou run open LLMs on Together already and want video on the same bill.
Each listing.
The directory page, alternatives in the same category, and a head-to-head where we have one.
- Together AIAlternatives
- falAlternatives
- ReplicateAlternatives
- Google VeoAlternatives
- Gemini Omni FlashAlternatives
- Runway APIAlternatives
- Luma APIAlternatives
- Kling APIAlternatives
- MiniMax Video (Hailuo)Alternatives
- Pika APIAlternatives
- Alibaba WanAlternatives
Category pages list every tool in that category, including ones left off this page. Built one of these? Get it featured in its category.
The 2 biggest AI stories, every morning.
One short email: the two stories that matter, and why.
You're in.
The Daily AI Brief will land in your inbox soon. If you don't see it, check Promotions or Spam and move it to your main inbox.