MiniMax H3 Video Model Hits Runware with Native Synced Audio
aziz4ai · x · 2026-07-31
MiniMax's new open-weights flagship video model, MiniMax H3, is now available on the Runware platform.
Key features of the model include:
- Native synced audio: Generates synchronized audio in the exact same pass as the video generation.
- Multi-reference conditioning: Maintains subject and scene identity consistency across multiple shots.
- Clip extension: Supports extending existing video clips or audio segments.
More from Multimodal
- Fan animation made in Blender with Claude Code (Opus 5.5) — Icy-Appointment3583 · 2026-10-16
- Making Blender animations with Opus, Astra and Argon: how you prompt still matters — chongdashu · 2026-10-16
- Tessellation-free geometric queries on displaced surfaces cut caustics variance 10x — ssh4net · 2026-10-16
- Complex multi-link CAD mechanism finally animates, dev celebrates the breakthrough — zetalyrae · 2026-10-16
- OpenMOSS releases MOSS-Image-SFT: Apache 2.0 text-to-image model on Hugging Face — bdsqlsz · 2026-10-16
- Hyperflow open-sourced: 49 sampling steps down to 8, ~3x faster video generation — mhdfaran · 2026-10-16