DeepSeek released DeepInfer under Apache 2.0, the framework it uses to serve DeepSeek-V4.1-Flash in production. It targets NVIDIA SM100 GPUs and Huawei Ascend 950 NPUs and currently supports offline inference only.
IBM and other contributors released version 1.0 of llm-d, an open-source distributed inference framework, ahead of PyTorch Conference North America. The release adds operational controls to help enterprises run large models reliably on their own infrastructure.
Interconnects' regular roundup of open models covers DeepSeek V4.1 Flash 1017, new American-made models and fresh regulatory controls in the US and China. The author frames it as Western open models striking back.
OVHcloud's boss revealed that the French cloud provider is developing its own large language model. A first version could launch at the end of November.
PS5LM Plus, an open-source project, uses the PlayStation 5's GPU to run AI models locally at more than 100 tokens per second. It only works on modified consoles.
Hugging Face published a post about TorchTPU, presented as a way to use TPUs natively. The headline signals support for the backend in the Hugging Face ecosystem.