Hugging Face Launches Inference Providers Hub and Multimodal Sentence Transformers v5.4; Arcee Releases 400B Trinity Large Thinking Model

Hugging Face introduced the Inference Providers Hub, aggregating endpoints from Together AI, Fireworks, Replicate, and others into a single model card interface covering 5,000+ models, allowing developers to benchmark latency and cost and switch providers with a single API parameter change — drawing praise as 'the missing abstraction layer' from ML engineers. Separately, Hugging Face released Sentence Transformers v5.4 with multimodal embedding and reranker models supporting text, image, audio, and video in a unified API for cross-modal search and multimodal RAG pipelines. Also debuting on Hugging Face: Arcee's Trinity Large Thinking, a 400-billion parameter open-weight reasoning model built by a 26-person team on a $20 million budget that CEO Mark McQuade claims is the most capable open-weight model from a non-Chinese company, directly challenging Chinese open-source dominance.