Newshot

US AI music giant moves into video generation, with underlying model from China's MiniMax

AI summary

US AI music platform Suno announced the launch of Visuals, a visual creation tool.

Translation(model translation, not the source's own text)

US AI Music Giant Moves into Video Generation, Underlying Model from China's MiniMax

US AI Music Giant Moves into Video Generation, Underlying Model from China's MiniMax

US AI music platform Suno announces the launch of visual creation tool Visuals

Recently, US AI music platform Suno announced the launch of its visual creation tool Visuals, bringing video and image generation into the music creation workflow. Suno officially disclosed that the model used by Visuals is called H3 Move, built on the open-source video model H3 from Chinese AI company MiniMax and tuned for music creation scenarios.

Suno Chief Product Officer Jack Brody said the Suno team optimized this open-source model using its own data to make it better suited to music video production. MiniMax H3 provides the underlying model foundation, and Suno builds on it with music-scenario adaptation and product integration, bringing song creation, visual generation, and character lip-sync into a single creative workflow.

According to Suno, videos generated by Visuals can echo the song's structure and lyrics, and support singing lip-sync. Users can choose from original characters offered by the platform or upload reference images, and combine music and text prompts to generate video clips of up to 15 seconds, then extend, arrange, and edit the clips via a timeline. In addition, users can create lyric videos and album covers, customize lyric fonts and colors, and choose aspect ratios according to publishing needs.

Video creation has always been a direction for Suno's product expansion, but in the past, slow generation speeds and high costs limited its use by ordinary users. Brody said video models have now reached the conditions in speed, cost, and quality to offer video creation services to ordinary users.

As the base model of H3 Move, MiniMax H3 opened its model weights in August this year, supporting enterprises and developers in deployment, fine-tuning, and customization on top of it. H3 can uniformly understand contexts composed of text, images, video, and audio, generate videos with native stereo audio, and support up to 2K resolution and 15-second duration.

Looking at the innovative product form of Suno this time, this kind of cross-modal capability naturally fits the creative needs of music videos. H3 Move presents one path for combining a base video model with a vertical application: supported by the base model's multimodal generation capabilities, then adding scenario optimization, editing features, and content protection mechanisms around music content.

Visuals uses an independent paid credits system. Video and image generation consume Visual credits, billed separately from song generation credits.

On content protection, Suno said Visuals incorporates text and visual filtering mechanisms, setting restrictions on specific artists, public figures' likenesses, and copyrighted content, and will embed watermarks before generated content is shared externally. According to Suno's official website, more than 100 million people have used its platform to create music.

This article is provided by MiniMax; QbitAI is authorized to repost it, and the views belong to the original author.

QbitAI's friends
QbitAI's friends
  • Sina News · 2026 Exploration Conference Held in Beijing: Blending Virtual and Real to Explore a New Vision of Symbiosis in the Intelligent Era 2026-11-20
  • Media Storm, Manus, Insta360, OPPO... GeekPark Innovation Conference 2027 Guest Lineup Updated! 2026-11-19
  • SenseNova U1 Pro Wins Double First in China and Sixth Globally on Design Arena 2026-11-19
  • The AI Infrastructure Wave Is Just Beginning, and Lenovo Group's Best Quarter Is Born! 2026-11-19

Related Reading

Hands-on with the Chinese product most like ChatGPT: It can explain The Three-Body Problem and gets memes from the "Retarded Bar"—beta open for everyone to try!

MiniMax Valued at HK$46.1 Billion, Raising HK$4.6 Billion, to Ring the Bell on January 9 Under Ticker 00100

Hangzhou's Reign Atop Global Open-Source Large Model Rankings Ends as Shanghai's Minimax M2 Sells Out Immediately Upon Release, a Million Tokens for Only 8 RMB

Breaking the impossible triangle of intelligence, cost, and speed

MiniMax Open-Sources New 4M Ultra-Long Context Model! Performance Rivals DeepSeek-v3 and GPT-4o

NVIDIA Team Builds Open-Source Video Enhancement Solution Based on MiniMax H3, Supporting Up to 4K Regeneration

The NVIDIA and MIT team released the model weights of the video enhancement project Open-H3-Regenerate-2K on Hugging Face

The World's First Large Model Earnings Report! MiniMax Foresees Three Major Super PMFs for 2026, and the AI Platform Company Sets Off

Full-year revenue of US$79.038 million

The translation is machine-generated and individual terms may be off. Nothing is rewritten or cut; the original is authoritative.

Published by 量子位
This site only summarises and aggregates; copyright stays with the original author. Headlines and summaries follow the source — where they differ, the original wins.
Read the original ↗