Tag
OpenInfer 0.2.0 released, supporting speculative decoding, p/d separation, and GLM5.2 FP8 inference; this project is a pure Rust + CUDA LLM inference engine with no dependencies on frameworks like PyTorch.