TorchTPU is Now Available

Native PyTorch on Cloud TPU: debug in eager mode, speed up with torch.compile, scale from one chip to a slice, with the ecosystem you already use

torchtpu

Foundations for TPU development

Build with your preferred framework and diagnostic tools to drive peak performance

Accelerate production inference on TPUs

Deploy high-throughput, low-latency workloads using vLLM and optimized TPU serving stacks

Scale model pre-training on TPUs

Achieve higher training throughput using JAX, PyTorch, and Keras on TPUs

Optimize post-training on TPUs

Efficiently customize and align open models for high-performance serving and deployment

TPU workload orchestration

Simplify cluster management and scale your workloads efficiently

Profile and debug TPU workloads

Uncover bottlenecks and optimize your model’s execution

Secure and scale TPU workloads

Design secure, high-performance network architectures for your AI workloads 

Community contributions

Resources and code samples from our developer community