跳到主要内容

AI 模型 API 目录

19 家厂商的 125 个模型系列,一个 API Key 全部调用。对比输入与实时价格,再在 Playground 或代码中运行。

Layout

24 个模型

清除筛选

30 s

longest clip

Kling

Kling 4.0

Generate video from a text prompt.

  • 图生视频
  • 文生视频
正在加载价格…
2 个版本

8 s

longest clip

Google

Veo 3.1

Google's flagship advanced AI Text-to-Video model, Veo3.1 Premium mode. Enables native text-to-video with fully synchronized ambient sound, dialogue and music, supports dialogue lip-sync, subject consistency and video interpolation. Generates top-tier cinematic videos with natural lighting, smooth camera transitions and strong narrative consistency, full flagship functions for professional storytelling and marketing, ultra-premium quality with ultra-high pricing.

  • 图生视频
  • 文生视频
正在加载价格…
9 个版本

15 s

longest clip

ByteDance

Seedance 2.0

Generate video using Seedance 2.0.

  • Reference to video
  • 图生视频
  • 文生视频
正在加载价格…
9 个版本

10 s

longest clip

MiniMax

Hailuo 02

Generate video from a text prompt.

  • 文生视频
  • 图生视频
正在加载价格…
7 个版本

16 s

longest clip

Vidu

Vidu Q4 Preview

Generate video with Vidu Q4 Preview.

  • Reference to video
  • 图生视频
正在加载价格…
2 个版本

10 s

longest clip

Google

Gemini Omni Flash

Gemini Omni Flash is a unified video generation model that creates high-quality short videos from text prompts. It supports 720p, 1080p, and 4k outputs with 4, 6, 8, and 10-second durations, making it suitable for creative clips, advertising assets, social videos, and visual concept demos.

  • 图生视频
  • 文生视频
  • Video edit
正在加载价格…
3 个版本

15 s

longest clip

xAI

Grok Imagine 1.5 Video

Leveraging xAI’s advanced reasoning capabilities, it excels at translating intricate prompts into visually stunning, logically coherent narratives. The model demonstrates superior performance in managing cinematic camera movements and complex physical interactions, ensuring that every frame adheres to realistic physics and lighting. It is designed for creators who demand high-precision storytelling and unparalleled visual fidelity without semantic loss.

  • 文生视频
  • 图生视频
正在加载价格…
2 个版本

10 s

longest clip

MiniMax

Hailuo 2.3

Generate video from a text prompt.

  • 文生视频
  • 图生视频
正在加载价格…
6 个版本

15 s

longest clip

Kling

Kling 3.0

Generate video from text with optional multi-shot scenes.

  • 图生视频
  • 文生视频
  • Video motion control
正在加载价格…
11 个版本

15 s

longest clip

Alibaba

Wan 2.7

Generate video from text with optional audio.

  • 文生图
  • Image tools
  • 文生视频
  • +3
正在加载价格…
9 个版本

15 s

longest clip

Alibaba

Wan 2.6

Generate video with Wan 2.6.

  • 文生视频
  • 图生视频
  • Reference to video
正在加载价格…
5 个版本

15 s

longest clip

Alibaba

HappyHorse 1.1

Generate video with HappyHorse 1.1 / Text to Video.

  • 文生视频
  • 图生视频
  • Reference to video
正在加载价格…
3 个版本

15 s

longest clip

Alibaba

HappyHorse 1.0

Generate video with HappyHorse 1.0 / Text to Video.

  • Reference to video
  • 图生视频
  • 文生视频
正在加载价格…
3 个版本

16 s

longest clip

Vidu

Vidu Q3

Generate video with Vidu Q3.

  • Reference to video
  • 图生视频
  • 文生视频
正在加载价格…
10 个版本

15 s

longest clip

MiniMax

MiniMax H3

Generate video using MiniMax H3.

  • 图生视频
  • 文生视频
  • Video to video
  • +1
正在加载价格…
9 个版本

15 s

longest clip

PixVerse

PixVerse V6

Generate video using PixVerse V6.

  • Video extend
  • 图生视频
  • 文生视频
正在加载价格…
4 个版本

10 s

longest clip

Kling

Kling 2.6

Generate video using Kling 2.6.

  • 文生视频
  • 图生视频
  • Video motion control
正在加载价格…
6 个版本

15 s

longest clip

Kunlun Wanwei

SkyReels V4

Generate video using SkyReels V4.

  • 图生视频
  • 文生视频
正在加载价格…
4 个版本

30 s

longest clip

Alibaba

Wan 3.0

Generate video with Wan 3.0.

  • 文生视频
  • 图生视频
正在加载价格…
4 个版本

10 s

longest clip

Kling

Kling O1

Generate video with Kling O1.

  • 文生视频
  • 图生视频
  • Video edit
  • +1
正在加载价格…
5 个版本

15 s

longest clip

Kling

Kling O3

Generate video with Kling O3.

  • 文生视频
  • 图生视频
  • Reference to video
  • +1
正在加载价格…
11 个版本

10 s

longest clip

Alibaba

Wan 2.5 Preview

Create a 5- or 10-second Wan 2.5 Preview video with optional audio input.

  • 图生图
  • 文生图
  • 图生视频
  • +1
正在加载价格…
4 个版本

20 s

longest clip

Black Forest Labs

FLUX 3 Video

Generate or continue a video with optional audio and draft mode.

  • 文生视频
  • 图生视频
  • Video to video
正在加载价格…
3 个版本

10 s

clip length

Meta

Muse Video

Generate visual content from a prompt and optional reference images.

  • 文生视频
正在加载价格…
1 variant

Video generation APIs

22 video model families from 11 providers. 21 turn text into video and 21 animate an image; clip lengths run from 1 to 30 seconds depending on the variant, and 7 families can generate sound with the video. Every request is an asynchronous job: submit, then poll.

Compare every price阅读快速开始

选择模型的常见问题

How do I call these models?

Image, video and music models are asynchronous: POST /api/v1/jobs with a model ID and an input object, then poll GET /api/v1/jobs/{id} until the task succeeds. LLMs are synchronous through POST /api/v1/chat/completions, which works with any OpenAI SDK. The quickstart has a request you can copy.

How am I charged?

From prepaid credits, with no subscription. A task's price is reserved when it is accepted and charged only when it succeeds; a failed task returns the reservation. LLMs are charged per token from the usage the model reports. Every rate is on the pricing page.

What is a model variant?

One model page can hold several model IDs: tiers such as Pro and Standard, tasks such as text to video and image to video, or a lower-priced channel. Each variant has its own model ID, input schema and price; pick one on the model page.

Which video models accept a start and an end frame?

Veo 3.1, Hailuo 02, Kling 4.0, Kling 3.0, Vidu Q3, MiniMax H3, PixVerse V6, Seedance 2.0, Wan 3.0, Kling O1, and 1 more. The variants table on each model page shows which variant takes a last frame.

Which video models generate sound?

Kling 3.0, Wan 2.6, PixVerse V6, SkyReels V4, Seedance 2.0, Kling O3 和 FLUX 3 Video. Sound is a field of the request, and it can change the price.

Why does a model say Price not set yet?

No price is configured for it yet, so it cannot run: the API answers MODEL_PRICE_UNAVAILABLE. The page and its schema stay available for reference.