The core of Atlas AI's basic software and hardware platform, designed to deliver ultimate performance, intuitive development, and an open ecosystem.

Community-driven, Atlas-adapted PyTorch that supports mainstream third-party ecosystems.

Fully compatible with the mainstream TensorFlow framework, enabling accelerated execution of TensorFlow networks on the Atlas platform.

An open, Atlas-adapted, multi-processor AI architecture for easy development, efficient running, and flexible deployment in all scenarios.

A distributed training acceleration library for Atlas hardware that leverages a rich set of algorithms and models to enable end-to-end acceleration.

An Atlas inference engine that enables efficient application deployment through hardware-based acceleration, fine-tuning, and rapid porting.

Enables full-lifecycle cluster management, including job scheduling, O&M monitoring, and fault recovery.

A full-stack toolchain providing end-to-end solutions for Atlas AI application development.

Consolidates industry capabilities to enable minimalist development and ultimate performance for industries.

A secure, lightweight device-edge-cloud synergy framework for managing edge AI services and building collaborative platforms.