PixVerse/v5

PixVerse V5 was officially released on August 27, 2025. Its core breakthrough is that the generation speed is compressed from "minute level" to "second level": users can generate a 360P short film in as fast as 5 seconds, and a 1080P high-definition video in 1 minute, with stable image quality and motion. Through technologies such as extreme distillation, human preference fitting, and unified feature space, the model has achieved significant improvements in light and shadow realism and aesthetic matching of complex character movements, surreal scenes, and real-life scenes. Larger-scale parameter design has greatly enhanced the model's ability to fit complex actions. The Agent creation assistant that is launched simultaneously is designed for users with no basic knowledge. When users upload a picture, it can automatically identify the characteristics and generate a complete short video of 5 to 30 seconds without mastering complex prompt word skills. V5 marks PixVerse’s transformation from a tool-based product to an intelligent creation platform.

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

Pricing

Price: $0.005 / request

360p : $0.225 /

360p : $0.45 /

540p : $0.225 /

540p : $0.45 /

720p : $0.3 /

720p : $0.6 /

1080p : $0.6 /

1080p : $1.2 /

Endpoints

  • openai

More models

  • veo-3.1-generate-preview

    veo-3.1-generate-preview 是 Google Veo 3.1 的视频生成预览模型条目,适合将故事脚本、镜头描述或视觉创意转化为短视频原型。其价值在于帮助团队快速验证叙事节奏、场景氛围、动作表现和镜头构图,再进入后续剪辑或制作环节。高质量提示应包含主体特征、动作顺序、镜头景别、光线、风格、时长和禁用内容。应建立包含人物、运动、文字、复杂场景和不同画幅的样片集来评测,并在外部传播前完成版权、肖像、商标和安全审核。

  • dreamina-seedance-2-0-fast-260128

    dreamina-seedance-2-0-fast 是 Seedance 2.0 的速度优化版本,继承了核心功能的同时大幅提升生成速度。它具备原生音频同步、摄像机控制和稳定动作能力,每次渲染成本更低,适合对生成速度有较高要求的电影剪辑场景。输出分辨率限制为 480P 和 720P(不支持 1080P)。在 Seedance 2.0 的产品矩阵中,Fast 版位于标准版和 Mini 版之间,是兼顾速度与质量的中档选择。

  • wan2.7-t2v

    Wan2.7-T2V 是文生视频模型,仅以文本提示为输入,直接生成带同步音频的视频内容。其核心定位是从零创作,模型能够根据简短的文字描述智能编排剧情节奏、设计镜头语言并安排场景转换,支持数十种基础运镜和复杂组合运镜方案。演绎能力全面升级,文戏情感细腻自然,动作戏激烈富有张力。输出支持720P和1080P分辨率,视频时长2至15秒,帧率30fps。该模型与系列中其他模型共享全模态输入体系,可通过阿里云百炼平台调用,是万相2.7“从演迈向导”创作链路中负责从零生成的起点环节。

  • wan2.7-i2v

    Wan2.7-I2V 是万相2.7视频系列中的图生视频模型,以图像为起点生成动态视频。其核心升级在于“演绎能力”的全面强化:文戏情感细腻自然,动作戏激烈流畅,镜头切换更具戏剧性和节奏感,能够实现更强的表演效果。模型支持首帧和首尾帧控制,可指定视频的起始与结束画面,同时支持片段续写,并能与音频同步生成对口型表演。输入模态涵盖图像、音频和文本,输出720P或1080P、时长2至15秒的视频,帧率30fps。