비디오 모델

Alibaba

Wan 2.5 is Alibaba's advanced video generation model featuring one-pass audio/video synchronization, multilingual support, and cost-effective production. Creates fully synchronized videos with voiceover and lip-sync from a single prompt.

Google

Google DeepMind’s upgraded AI video model for realistic motion generation, extended clip duration, multi-image reference control, and synchronized audio output in native 1080p.

OpenAI

Sora 2 is OpenAI’s latest AI video generation model, supporting both text-to-video and image-to-video. It delivers realistic motion, physics consistency, with improved control over style, scene, and aspect ratio—ideal for creative apps and social media content.

OpenAI

Sora 2 Pro is OpenAI’s premium video generation model with higher quality output, supporting text-to-video and image-to-video at 720p and 1080p resolutions with flexible durations of 10 or 15 seconds.

bytedance

Seedance 1.5 Pro is ByteDance's per-second video model for fast text-to-video and image-to-video generation with 480p/720p output, optional sound, aspect ratio control, and fixed-lens camera stability.

vidu

Vidu Q3 is a per-second video generation model that combines standard and Turbo text-to-video plus image-to-video workflows in one API. It supports single-image animation, first-and-last-frame transitions, optional sound and BGM, and output up to 1080p.

kling

Kling 2.1 is Kuaishou's multi-tier video model with Standard, Pro, and Master modes for image-to-video and text-to-video creation. It supports 5-10 second clips, optional tail images for Pro, and aspect ratio control for Master text-to-video.

kling

Kling 2.5 Turbo Pro is Kuaishou's high-speed video model for text-to-video and image-to-video creation. It supports 5-10 second clips, optional tail-frame images, aspect ratio control for text-to-video, plus negative prompts and CFG scale guidance.

kling

Kling 2.6 is Kuaishou's native audio-visual video model that generates video, speech, sound effects, and ambience in one pass. It supports text-to-audio-visual and image-to-audio-visual creation with Chinese and English voice generation and up to 10-second clips.

kling

Kling 3.0 Std is Kuaishou's standard-quality video generation model with text-to-video, image-to-video, and motion-control modes. It supports clips up to 15 seconds with optional sound generation and flexible aspect ratios.

Lightricks

LTX-2 19B is Lightricks' open-source 19B diffusion transformer for cinematic video generation. It supports text-to-video and image-to-video workflows, LoRA conditioning, and high-fidelity outputs up to 1080p in the API.

wavespeed

InfiniteTalk converts one photo plus audio into audio-driven talking or singing avatar videos with precise lip synchronization. Supports up to 10 minutes at 480p or 720p resolution.