Hugging Face has just released version v1.16.0 of its core library, focusing heavily on multimodal inference capabilities with support from its partner Together Compute.
Key Developments
This update establishes Together Compute as an official Inference Provider, supporting five new task types: feature extraction, text-to-speech (TTS), automatic speech recognition (ASR), image-to-image, and notably, text-to-video generation.
Why It Matters
Integrating high-performance providers like Together Compute reduces the Hugging Face ecosystem's reliance on a single source while lowering the barrier to accessing heavy models like video generation. This is great news for Vietnamese developers building multimodal AI applications that require fast processing speeds and optimized costs.