Models are tools inside a workflow—not interchangeable badges.

Each entry distinguishes what the model developer documents, what MagicLight currently displays, how a creator can test it, and what remains unknown. Last checked August 27, 2026.

ByteDance Seed

Seedream 5.0 Pro

An official-source guide to Seedream 5.0 Pro for MagicLight creators, covering structured images, text rendering, storyboard workflow, prompts, limitations, and image-to-video handoff.

Structured image generation for storyboards →OpenAI

GPT Image 2

A verified GPT Image 2 guide for MagicLight creators covering text-to-image, image editing, storyboard visuals, diagrams, prompts, limitations, access, and rights checks.

Image generation and editing for storyboards →ByteDance Seed

Seedance 2.5

A verified guide to Seedance 2.5 inside MagicLight, covering 30-second audio-video generation, references, editing, prompts, limits, and story use cases.

30-second audio-video storytelling →Kuaishou

Kling AI 3.0

A practical guide to Kling AI 3.0 in MagicLight, including model variants, multimodal workflows, native audio, duration, prompts and limitations.

Multimodal video and image model series →Google DeepMind

Veo 3.1

A verified Veo 3.1 guide for MagicLight users covering text-to-video, image-to-video, audio, reference ingredients, workflow checks, and limits.

Cinematic video with audio and references →OpenAI

Sora 2

A careful guide to the Sora 2 name shown by MagicLight, its documented capabilities, safety context, and OpenAI's current product-availability caveat.

Historical video-and-audio model availability guide →