$ the-wire · showcase
OpenRouter joins Inference Providers in huggingface.js
By RepoJournal · Filed · About Hugging Face · Composed from the cited sources · methodology
huggingface.js adds OpenRouter as a provider at the helper layer, which is where Hub provider requests, mapping validation, and the model page widget all resolve, so the provider works across the Hub rather than in one surface.
Add OpenRouter as a provider (#2639) huggingface/huggingface.js
OpenRouter is wired in as a chat-completion provider, and because the Hub routes provider requests, mapping validation, and the model page widget through huggingface.js helpers, it lands everywhere a provider normally appears. The contributor notes the change arrived as an agent-authored pull request (Opus 5.5) reviewed by @hanouticelina, with other task types to follow the same path later.
inference-v4.13.32 huggingface/huggingface.js
The inference package cut inference-v4.13.32, the release that carries the OpenRouter provider work to consumers. Corresponding tag commit published as @huggingface/inference 4.13.32.
Enable native writing assistance in mobile Reader and Overview huggingface/agent-manager
Native autocorrection, sentence capitalization, and spellcheck are now always on in mobile Reader and Overview reply boxes, with no toggle and no stored preference to configure. Desktop and raw terminal input stay literal, and IME Enter/Escape handling is protected in both the Composer and the enclosing Overview window, including Safari's keyCode 229.
Add Service Accounts row to Enterprise plans comparison (#2874) huggingface/hub-docs
The Enterprise plans comparison table gains a Service Accounts row, useful if you are mapping plan tier to what your org can actually provision.
[Docs] Fix indentation in ZipScheduler examples huggingface/huggingface_hub
The ZipScheduler examples in the English, German, Hindi, and Korean upload guides had bad indentation: copying the existing definition raised IndentationError at the following archive block. Each corrected snippet now compiles. The bot-driven Inference Providers documentation regeneration and a README refresh in pytorch-image-models round out the day.
