Introducing ML 3.0: The Next Generation Platform
Announcing major updates to our platform including new algorithms, better GPU support, and enhanced collaboration features.
Alex Kim
VP of Product
Now Available
Today, we're thrilled to announce the general availability of ML 3.0—the most significant update to our platform since we launched. This release represents months of work based on feedback from our community of over 50,000 machine learning practitioners.
What's New in ML 3.0
New Algorithm Library
50+ new pre-built algorithms including state-of-the-art transformers, diffusion models, and RL agents.
2x Faster Training
Optimized kernels and improved GPU utilization deliver 2x faster training on average.
Team Workspaces
Collaborate seamlessly with shared experiments, datasets, and model registries.
Enterprise Security
SOC 2 Type II certified with new RBAC, audit logs, and private deployments.
Enhanced GPU Support
ML 3.0 introduces native support for NVIDIA H200 and AMD MI350X GPUs, along with optimized kernels that automatically select the best configuration for your hardware. Multi-node training is now seamless with our new distributed training coordinator.
Getting Started
Existing users are automatically upgraded to ML 3.0. New users can get started with our free tier, which now includes 100 GPU hours per month. Check out our updated documentation and tutorials to explore the new features.
Alex Kim
VP of Product
Alex leads product at 1.ML, previously Product Lead at Meta AI.