wan3.0-video-prime

Wan3.0-Video-Prime is a high-speed version of Wan3.0-Video, which greatly improves the end-to-end generation speed while maintaining a fully aligned capability system with the standard version. It also supports multi-modal input such as text, image, video, audio, etc., and uniformly realizes the capabilities of text video, image video (first frame/first and last frame) and reference video. The output resolution covers 480P, 720P and 1080P, and can generate videos of up to 30 seconds with a frame rate of 30fps. This model shares the same set of APIs with the standard version, and developers can switch between them without modifying the code. It is suitable for mass production scenarios that require higher generation speed. The two are fully aligned in terms of capabilities, and the Prime version meets real-time or near-real-time video generation needs with faster inference speed.

Context
0 tokens
Max output
0 tokens
Input
llm image video
Output
video

Pricing

Price: $0.28 / request

480P reference_to_video : $0.07 /

480P reference_to_video : $0.07 /

720P reference_to_video : $0.14 /

720P reference_to_video : $0.14 /

1080P reference_to_video : $0.28 /

1080P reference_to_video : $0.28 /

Endpoints

  • openai

More models

  • wan2.7-videoedit

    Wan2.7-VideoEdit 是视频编辑模型,用户通过自然语言指令即可对已有视频进行局部或全局修改。编辑能力覆盖多个维度:可添加、删除或替换视频元素,切换场景环境(如晴天改雨天),变换整体视觉风格(如写实转水彩、真人转黏土动画);角色行为、表情和情绪均可调整,台词可替换并自动匹配口型与音色;拍摄方式也可重新定义,包括机位、视角、景别和焦距等参数。模型还支持参考图像替换视频元素,并复刻视频中的动作、特效和运镜等动态过程。输入模态为图像、文本和视频,输出720P或1080P视频,时长2至10秒

  • veo-3.1-lite-generate-preview

    veo-3.1-lite-generate-preview 是 Google Veo 3.1 的轻量视频生成预览模型条目,适合成本与迭代速度优先的视频原型工作流。它可用于大量探索不同的创意方向,例如为同一产品卖点生成多种分镜、背景、动作或短片版本,再筛选出值得进一步精修的方案。提示词应聚焦一个清晰主体和核心动作,避免在单次生成中堆叠过多镜头变化。建议以同一批脚本比较其质量、生成时延和可控性,并将涉及真实人物、受保护角色、商标或音乐的内容交由人工审核。

  • MiniMax-H3

    MiniMax-H3是一款通用的全模态生成模型,支持对文本、图像、视频、声音组成的多模态上下文的统一理解能力、能够输出具备原生双声道的音视频,最高可支持 15s 2K 分辨率。

  • happyhorse-1.1-i2v