Tencent's video model demo now runs locally: 7.5GB encoder cut and an MLX branch
gaganghotra_ · x · 2026-09-18
Tencent's open video generation model demo now runs on your own GPU. After launching needing 25GB of VRAM, the week brought CPU offload support, a 7.5GB encoder size reduction, and an MLX branch for Apple Silicon. The poster quips that the ecosystem 'just keeps on accelerating,' underscoring how fast open video models are getting leaner for local deployment.
More from Infra
- Google, Nvidia and Anthropic want to unlock 100 GW for AI — power may be the real bottleneck — TansuYegen · 2026-09-18
- Huawei Moves Ascend 960DT AI Chip Launch Up to Early 2027 — emmanuelvivier · 2026-09-18
- Subsidised coding subscriptions will shrink; policy-based token routing is coming — craigbalding · 2026-09-18
- Flyweight: open-source C++/CUDA engine runs VRAM-busting MoE models on one GPU plus system RAM — Main-Wolverine-1042 · 2026-09-18
- AI ported a distro in 20 minutes — is NVIDIA sawing off its own CUDA moat? — BringTea_666 · 2026-09-18
- India Lands $12 Billion in Semiconductor Investment Pledges Within Months — pstAsiatech · 2026-09-18