Lightx2v Releases MiniMax H3 Turbo Ref2V Video Model
welt101 · reddit · 2026-08-13
Lightx2v has released the MiniMax H3 Turbo Ref2V model on Hugging Face. The page is now live, allowing users to directly access the weights for this video generation model.
Based on the naming, this version likely adds Ref2V (Reference-to-Video) capabilities on top of Turbo acceleration.
More from Multimodal
- Flux 3 Klien first impressions: occasional body horror like SD3, otherwise solid — rionix88 · 2026-10-16
- XGRIDS builds world's largest 3DGS town: 23.5 km² and 6.2 billion Gaussian Splats — bilawalsidhu · 2026-10-16
- IMMERSION: AI film compilation links 12 shorts with one continuous camera move — AI_Andrew · 2026-10-16
- A song with vocals, lyrics and 3D head animation created entirely by Opus 5.5 — repligate · 2026-10-16
- Alibaba open-sources CommerceGen-Studio for commercial-grade e-commerce images — alibabagroup · 2026-10-16
- EVA-Cap: Fudan and OPPO align video captioning with atomic audiovisual events via GRPO — Fudan-University · 2026-10-16