PixVerse/v5.6

PixVerse V5.6 was released on January 26, 2026. The core upgrade is "amazing physical simulation" capabilities and movie-level picture generation. The model has achieved significant progress in physical motion simulation, and the movement trajectories of objects and characters are more in line with the mechanics of the real world. Enhancements have also been made to lip sync and audio realism. V5.6 supports the unified model of Vincent Video and Tusheng Video, with resolutions ranging from 360p to 1080p, and duration supporting 5 seconds, 8 seconds and 10 seconds. 10-second videos are only available in 360p, 540p and 720p resolutions, and 1080p is limited to 5 or 8 seconds. The model also supports first and last frame transition generation and prompt word reasoning enhancement. In the global Tusheng video track of Artificial Analysis in February 2026, V5.6 ranked fourth with an ELO comprehensive quality score of 1291.

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

Pricing

Price: $0.005 / request

360p without : $0.17500000000000002 /

360p without : $0.35000000000000003 /

360p without : $0.385 /

540p without : $0.17500000000000002 /

540p without : $0.35000000000000003 /

540p without : $0.385 /

720p without : $0.225 /

720p without : $0.45 /

720p without : $0.495 /

1080p without : $0.375 /

1080p without : $0.75 /

360p with : $0.4 /

360p with : $0.5750000000000001 /

360p with : $0.61 /

540p with : $0.4 /

540p with : $0.5750000000000001 /

540p with : $0.61 /

720p with : $0.4 /

720p with : $0.675 /

720p with : $0.72 /

1080p with : $0.75 /

1080p with : $0.975 /

Endpoints

  • openai

More models

  • dreamina-seedance-2-5-260628

    dreamina-seedance-2-5-260628 是 Dreamina Seedance 系列的最新一代视频生成模型,于 2026 年 8 月正式上线。它支持文生视频和图生视频工作流,输入模态丰富,包括图片、视频、音频以及预上传的虚拟形象素材,支持首帧和尾帧锚定、4-15 秒灵活时长、480P 和 720P 输出分辨率、多种画面比例以及可选音频合成。模型还支持联网搜索增强生成,可通过实时网络检索获取最新信息来辅助视频创作,并支持最多 9 张参考图片、3 段参考视频和 3 段参考音频。Seedance 2.5 在 2.0 基础上进一步强化了多模态参考能力和生成质量。

  • PixVerse/v5.5

    PixVerse V5.5 于2025年12月1日发布,是首款实现“分镜+音频”一键直出的 AI 视频大模型。此次更新首次支持音频与多镜头同步生成,并强化了多角色音画同步能力,AI 能够根据提示词生成具备叙事结构的短片,标志着 AI 视频从“镜头生成”向自动“讲故事”的进化。用户可在多镜头和音画同步开启状态下生成5秒、8秒或10秒的视频,直接在提示词中控制音效类型。在分辨率方面,10秒片段最高支持720p,1080p输出限制为5秒或8秒。V5.5新增了 generate_audio_switch 和 generate_multi_clip_switch 等 API 参数,为开发者提供了更细粒度的控制能力。

  • happyhorse-1.1-t2v
  • happyhorse-1.0-t2v

    HappyHorse-1.0-T2V 是文生视频变体,仅以文本为输入,直接生成带同步音频的视频内容。模型具备“高度还原的动态画面生成能力”,能够精准理解文本语义,输出流畅自然、细节丰富的高质量视频。与 I2V 一样,T2V 的推理服务由阿里云百炼提供,支持全球多区域部署,RPM 同为 300。在 Artificial Analysis 的文生视频竞技场中,T2V 赛道以 Elo 1333 分登顶,超越了字节跳动 Seedance 2.0 等热门模型。该模型原生支持英语、普通话、粤语、日语、韩语、德语、法语七种语言的唇形同步,在单张 H100 上生成 5 秒 1080p 视频约需 38 秒。