Realtime AI News
Suno launches Visuals, building music video generation on MiniMax's open H3
Suno launched Visuals, a visual creation tool that folds video and image generation into its music workflow, powered by a model called H3 Move that is built on MiniMax's open-source H3 and tuned for music use. Users can generate clips of up to 15 seconds from music and text prompts, with singing lip sync, timeline editing and lyric videos.

The US AI music platform Suno has launched Visuals, a visual creation tool that brings video and image generation into its music-making workflow. According to Suno's disclosures, the model behind it is called H3 Move, built on the open-source video model H3 from Chinese company MiniMax and tuned for music-creation scenarios.
The division of labor is stated plainly. Suno chief product officer Jack Brody said the team optimized the open model with its own data to make it better suited to music videos, with MiniMax H3 supplying the base model and Suno handling scenario adaptation and product integration, so songwriting, image generation and lip sync sit in one pipeline.
On capabilities, Suno says Visuals output responds to a song's structure and lyrics and supports singing lip sync. Users pick from original characters provided by the platform or upload reference images, combine music and text prompts to generate clips of up to 15 seconds, then extend, arrange and edit them on a timeline. They can also produce lyric videos and album covers, customize lyric fonts and colors, and pick aspect ratios for different platforms.
The product rests on a cost judgment. Video has long been an obvious expansion path for Suno, but generation used to be slow and expensive, which kept ordinary users out. Brody's position is that video models now meet the bar on speed, cost and quality needed to offer video creation to mainstream users.
As the base, MiniMax H3 opened its weights in August, allowing companies and developers to deploy, fine-tune and customize it. Public descriptions say H3 understands a unified context of text, images, video and audio, generates video with native stereo audio, and supports resolutions up to 2K and clips up to 15 seconds.
That cross-modal reach fits music video needs naturally. H3 Move shows one path for pairing a base video model with a vertical application: lean on the foundation model's generation ability, then add scenario tuning, editing features and content protection around music content rather than asking the application layer to solve generation quality from scratch.
Monetization and compliance are handled separately. Visuals uses its own paid credit system, where video and image generation consume Visual credits billed apart from song generation credits. On protection, Suno says Visuals includes text and visual filters that restrict specific artists, public figures and copyrighted material, and embeds a watermark before generated content is shared externally. Suno's website states that more than 100 million people have created music on the platform.
The provenance matters. QbitAI labels the piece as supplied by MiniMax and republished with authorization, so the product details and executive statements reflect the MiniMax and Suno side of the story. Suno has not disclosed Visuals pricing tiers, availability by region or generation queue times, and those are the practical questions for whether ordinary users can actually use it.
Why it matters
It is a clear case of an open base model being adopted directly by a leading overseas app, showing that Chinese video models now compete on being embedded in commercial products rather than on benchmark numbers. For Suno, video also stretches its competitive surface from music generation into short-form creation tools.
Nearby Updates
All11/20, 21:03
Stripe founders and Paul Graham back agent-swarm startup Hoplite with $5.5M
Hoplite, a startup working on AI agent swarms, has raised $5.5 million from backers including the founders of Stripe and Paul Graham, according to an exclusive Sifted report. The public excerpt discloses no product details, round type or valuation, but the investor list points to interest in multi-agent orchestration.
11/20, 19:00
Alibaba Cloud adds a third Johor data center aimed at local AI workloads
Alibaba Cloud has added a third data center in Malaysia's Johor state, positioned to serve local AI workloads, according to Unite.AI. The addition points to continued capacity build-out in Southeast Asia rather than a one-off landing.
11/20, 21:57
Enterprise Autonomy launches an open benchmark for AI agent governance
Enterprise Autonomy announced an open benchmark for AI agent governance in a press release carried by EIN News. The announcement did not disclose the benchmark's scoring dimensions, maintainers or code location, so its usefulness depends on methodology that has yet to be published.
11/20, 18:00
OpenAI Lays Out a Four-Layer Defense Against Adversarial Distillation
OpenAI published a post on November 20 describing how it fights adversarial distillation by stacking model training, safeguards, detection and enforcement. The company frames capability extraction as a standing problem that no single filter can solve.