JAX

JAX is a Python library for accelerator-oriented array computation and program transformation, designed for high-performance numerical computing and large-scale machine learning. It automatically differentiates Python and NumPy functions, including through loops, branches, and recursion, and also supports higher-order derivatives. It compiles NumPy programs with XLA to run at scale on hardware accelerators such as TPUs and GPUs. Composable function transformations such as grad, jit, and vmap are at its core, and it states that it is a research project, not an official Google product.

• Automatic differentiation of Python and NumPy functions (reverse and forward mode, arbitrary order)
• XLA-based jit compilation
• Automatic vectorization with vmap
• Scaled execution on accelerators such as TPUs and GPUs
• Distributed arrays with automatic parallelization, and manual parallelization with shard_map
• Write custom TPU and GPU kernels with Pallas
• Foreign function interface (FFI)

Open source + paid cloud-managed plans

Open-source version can be self-hosted for free, managed cloud services are paid based on usage/seats

Estimate based on AI knowledge - please verify on the official site

• One of 45 AI tools in the MLOps & Training field.
• Connected to 5 other tools — #11 in its field.