Access powerful on-demand AI compute with Graphcore IPUs.

Accelerating machine learning workloads including NLP, Computer Vision and Graph Neural Networks.

Train, fine-tune and run inference on Gradient Notebooks on FREE Cloud IPUs.

Graphcore and Paperspace are working together to make training models on IPUs fast and easy.

Integrated with the frameworks you love

Graphcore’s Poplar® software stack is co-designed with the IPU for AI application development.

Poplar supports many popular machine learning frameworks including TensorFlow, PyTorch, Hugging Face, Keras, PaddlePaddle and more.

Perfect for ML developers. A production-ready development environment backed by Graphcore IPUs.

Easy setup
sample code
Free IPUs

AI compute on demand

Available globally with an unlimited number of six-hour free sessions, allowing ML developers and researchers to try IPUs without friction or any upfront costs.

Instant remote execution

Select an IPU runtime with pre-configured Docker containers and launch from your browser in seconds within Gradient's simple web-based Jupyter IDE.

Access high-end IPUs

Instantly access IPU-POD4, IPU-POD16 and Bow Pod16 machines delivering up to 5.6 petaFLOPS of AI compute, designed from the ground up for new breakthroughs in machine intelligence.

Run AI models out-the-box

Effortlessly run state-of-the-art ML models accelerated by the architectural advantages of the IPU - including NLP, Computer Vision, and Graph Neural Networks (GNNs).

Self-guided resources

Discover a range of self-guided learning content including hands-on tutorials and code examples from basic programming concepts to optimisation techniques.

ML framework support

Fully integrated with common ML frameworks including PyTorch and TensorFlow, plus popular libraries including Hugging Face Optimum transformers.

Testimonial Image

"IPU-powered Gradient Notebooks is a great way to discover the performance advantages of Graphcore IPUs in the cloud. The process is so easy - with 1 click in my browser, the team at Observatoire de Paris - PSL was able to explore a selection of out-the-box IPU-ready models covering CV, NLP & GNNs, with the simplicity of Paperspace's cloud development environment."

Damien Gratadour

Associate Professor

Ready to get started? Get in touch or create an account.