TorchServe GitHub repository (PyTorch Serve)

README carries an explicit notice: 'This project is no longer actively maintained. While existing releases remain available, there are no planned updates, bug fixes, new features, or security patches. Users should be aware that vulnerabilities may not be addressed.' A security announcement notes TorchServe now enforces token authorization enabled and model API control disabled by default, to address unauthorized API calls and the risk of malicious code being introduced to the model server. Self-described as 'a flexible and easy-to-use tool for serving and scaling PyTorch models in production', requires Python >= 3.8, installable via torchserve / torch-model-archiver / torch-workflow-archiver with nightly variants, with optional ROCm and CUDA accelerator flags. CI badge set reveals the real integration story: nightly build, docker nightly, benchmark nightly, docker regression, KServe regression and Kubernetes regression. API metadata 2026-10-04: created 2019-10-03, last push 2025-08-06, 4343 stars, 879 forks, 443 open issues, ARCHIVED (archived: true), Apache-2.0.