Purpose-Built for Your AI Solution Systems
Purpose: Instant use of your AI system without setup.
Preloaded: Conda environments (TensorFlow, PyTorch, RAPIDS).
Best For: Data scientists, labs, and rapid POCs.
Value: Begin model training on Day One.
Purpose: Containerized environments for hybrid deployment.
Includes: OCI containers + NVIDIA Container Toolkit.
Best For: ML Ops, CI/CD, and versioned deployments.
Value: Run the same container on dev, server, or cloud.
Purpose: Full control over the software stack.
Minimal Install: Driver + CUDA only.
Best For: Regulated, secure, or air-gapped setups.
Value: Compliance-ready and customizable by design.
| Category | Software Component | LaunchPad | FlexBox | ForgeKit |
|---|---|---|---|---|
| AI/ML Frameworks | TensorFlow V2 | Yes | Yes | No (User Add) |
| TensorFlow V1 | No | Yes | No | |
| PyTorch | Yes | Yes | No | |
| MXNet | Yes | Yes | No | |
| Caffe / Caffe2 | Yes (Caffe2) | Yes | No | |
| Chainer | No | Yes | No | |
| Microsoft CNTK | No | Yes | No | |
| Libraries | cuDNN | Yes | Yes | Yes |
| RAPIDS | Yes | Yes | No | |
| Keras | Yes | Yes | No | |
| Theano | Yes | Yes | No | |
| OpenCV | Yes | Yes | No | |
| Dev Tools | CUDA Toolkit | Yes | Yes | Yes |
| NVIDIA DIGITS | Yes | No | No | |
| Anaconda | Yes | Yes | No | |
| NVIDIA Container Toolkit | No | Yes | No | |
| Orchestration | MicroK8s | Optional | Optional | Optional |
| CNCF/Kubernetes | No | Optional | Optional | |
| Docker / Podman | No | Yes | Optional |