Sand.ai

Description

Sand.ai is an AI company specializing in video generation technology and products. It has independently developed Magi-1, an autoregressive video model, and GAGA-1 for simultaneous audio-visual generation, and released MAGI-2 Preview, described as a large-scale open-source Mixture-of-Experts video generation model.

MAGI-2 Preview is a 114B-parameter unified audio-video generation model that activates just 6B parameters per token. Built on MagiMoE and co-designed across architecture, systems, and data, it generates dialogue, singing, and emotional shifts with natural coordination of lip-sync, facial expressions, eye movements, and body language. Because visuals and audio are generated jointly by a single model, dialogue and lip movements, actions and sound effects, and environmental changes and audio feedback stay precisely aligned. The model also understands shot scale, composition, perspective, and camera-movement intent, producing motions such as zooming, tracking, and orbiting for a filming-like visual rhythm.

Developers can access Sand.ai's capabilities through its API Platform. After logging in with a Magi account, users can purchase credits, create API keys, and authenticate calls via an Authorization header. Platform and Magi billing are kept separate, so API usage does not consume Magi account credits.

Sand.ai's Core Features

• MAGI-2 Preview, a 114B-parameter unified audio-video generation model

• Joint audio and video generation with precise lip-sync alignment

• Character performance: dialogue, singing, and emotional shifts

• Cinematography control over shot scale, composition, and camera movement

• Magi-1 autoregressive video model and GAGA-1 audio-visual model

• Developer API platform with credits and API key management

How to use Sand.ai?

1. 1

   Create an account: Log in to the platform using your Magi account.

2. 2

   Buy credits: Purchase a credits package for API usage.

3. 3

   Create an API key: Generate a key in the API Key management page and save it securely.

4. 4

   Call the API: Authenticate with the key via the Authorization header to generate audio-video content.

Sand.ai's Use Cases

• Talking-character video
• Cinematic video generation
• Audio-visual content creation
• Developer integration

FAQ from Sand.ai

• What is Sand.ai?

  Sand.ai is an AI company focused on video generation, known for the Magi model family and the MAGI-2 Preview audio-video generation model.

• What is MAGI-2 Preview?

  It is a 114B-parameter unified audio-video generation model that activates about 6B parameters per token and generates synchronized visuals and audio.

• How does it keep audio and video in sync?

  Visuals and audio are generated jointly by a single model, aligning dialogue with lip movements, actions with sound effects, and scene changes with audio feedback.

• Are the models open source?

  MAGI-2 Preview was released as a fully open-source project, and Magi-1 is an autoregressive video model developed by the company.

• How do developers access Sand.ai?

  Through the API Platform: log in with a Magi account, buy credits, create an API key, and authenticate requests via the Authorization header.