生成片段，剪成一部影片。

Google's high-quality video model with native audio, reference-image support, and frame-to-frame control.

Google's fast video model for generating high-quality short clips from text prompts.

Google's cost-efficient Veo tier, generating short clips with native audio at roughly half the price of Veo 3.1 Fast.

Runway's flagship video model with realistic physics and strong visual fidelity for text-to-video and image-to-video.

Kuaishou's unified video model: clips of up to 30 seconds at 1080p with native audio, from text, first and last frames, reference images and videos, or a prompt that edits a source clip.

Kuaishou's faster, lower-cost Kling 4.0 tier: 720p clips of 3 to 20 seconds with native audio, from text, a first frame, reference images and videos, or a prompt that edits a source clip.

Kuaishou's cinematic video model generating clips up to 1080p with native audio, lip sync, and sound effects from text or images.

Google's updated omni video model, adding a 360p draft tier and 4K output, start-to-end keyframe interpolation, reference videos, and natively synchronized audio.

Google's omni video model, successor to 1.1: it generates, edits and extends clips of up to 15 seconds with natively synchronized audio, from text, start and end frames, reference images and videos, at 360p to 4K.

Black Forest Labs' multimodal video model: text, image, or video in, with synchronized audio, lip-sync, and a cheap draft mode.

Black Forest Labs' fast video editing model: rewrites a source clip from a text prompt, adding, removing or replacing objects and settings while keeping its length, aspect ratio and audio.

Pruna AI's video editing model: rewrites a source clip of up to 15 seconds from a text prompt, optionally guided by up to four reference images, with a low-cost draft mode.

Black Forest Labs' video upscaler: enlarges a source clip by 1.5 to 3 times, either staying faithful to the footage or adding fine detail, with an optional prompt to steer it.

Lightricks' high-fidelity video model — native synchronized audio, first-to-last-frame control, and directable camera moves at up to 1080p.

Lightricks' fast video model — synchronized audio, first-to-last-frame control, and clips up to 20 seconds, at resolutions up to 4K.

MiniMax's Hailuo 3 video model with native synchronized audio, multi-reference consistency, and prompt-driven editing of a finished clip.

MiniMax's throughput-tuned H3, built for faster clips with native synchronized audio and first-to-last frame control. It trades H3's multi-reference conditioning and 2K tier for speed.

A faster, lower-cost tier of MiniMax's H3 Max for text-to-video and first-to-last frame clips of 5 to 15 seconds with native audio, at 480P or 768P.

MiniMax's fastest H3 tier: 480p clips of 4 to 15 seconds with native audio from text, first and last frames, or reference images, videos and audio.

MiniMax's lower-cost H3 tier for drafts and high-volume video, at 480P or 768P, with first and last frames and image, video and audio references.

MiniMax's faster H3 tier for everyday video at 768P, balancing quality and speed, with first and last frames and image, video and audio references.

HeyGen's general-purpose video model: clips of 5 to 15 seconds with synthesized dialogue, ambience and sound effects, from text, a starting image or reference media, at 480p or 768p.

xAI's low-cost Grok Imagine video tier: clips of 1 to 15 seconds with synchronized audio from text or a starting image, at 480p to 1080p.

xAI's image-to-video model — animates a still image into a short clip with synchronized audio.

Alibaba's video model with three modes — text-to-video, image animation, and reference-to-video from up to 9 images.

ByteDance's flagship multimodal video model — native synchronized audio, clips up to 30 seconds, and large reference sets for character and scene consistency.

ByteDance's advanced video model with improved motion coherence and scene quality.

ByteDance's lightweight Seedance 2.0 tier — about half the cost per second, 480p/720p with synchronized audio.

ByteDance's low-latency Seedance 2.0 tier — quick 480p/720p generation with synchronized audio.

Alibaba's all-in-one Wan 3.0 video model, generating clips of up to 30 seconds with native audio from text, frame images, or a deep bench of reference media.

Alibaba's lower-latency Wan 3.0 tier, with the same multimodal workflows, 30-second ceiling, and native audio as the base model.

Pruna AI's quality-focused successor to P-Video, turning text, frame images, or a driving audio track into clips of up to 20 seconds with native audio and a low-cost draft mode.

Pruna AI's highest-quality video model, turning text or first and last frame images into 5 to 15 second clips at up to 768p with native audio, in a fast Speed mode or a slower Quality mode.

Runway's fast, low-cost Gen-4 image-to-video model: animates a start frame into a 2 to 10 second clip.

Runway's video editing model: restyles, relights, or changes a 2 to 30 second source clip from a text prompt, with an optional first-frame keyframe and outpainting to a new aspect ratio.

Kuaishou's faster Kling 3.0 tier, generating 3 to 15 second clips up to 1080p with native audio from text or a first frame.

Kuaishou's all-in-one Kling 3.0 model that generates or edits clips up to 4K from text, frames, reference images and a source video.

Alibaba's compact 5B video model, running fully offline on your own machine. Handles both text-to-video and image-to-video, and needs no API key or network once the weights are downloaded.

Alibaba's smallest Wan video model, and the lightest local option. Runs at 480p on modest hardware and shares its text encoder with the other Wan entries, so it costs little extra once one of them is installed.

Alibaba's largest open Wan model, a two-expert mixture that swaps between a high-noise and a low-noise transformer partway through denoising. Highest quality of the local Wan entries and by far the heaviest.

Lightricks' 22B video model, distilled for fewer sampling steps. Generates synchronised audio alongside the video, and drives a Gemma 3 language model as its text encoder.

MiniMax's H3 video model running offline, with synchronised audio. The heaviest local entry by a wide margin: its text encoder alone is a 32B vision-language model.