Inference Stack

Know what actually runs together before you build it

Trace a real path from your use case to a cloud instance — across NVIDIA, AMD, Google TPU, AWS Trainium, and Intel Gaudi — and see exactly what's compatible at every layer, not just what's available.

See reference architecturesBrowse the catalog
7 layers50 components5 silicon vendors
Trace it
Start from what you're building. See every compatible framework, orchestration layer, and silicon path — not a single vendor's funnel.
Ground it
Don't guess your stack. See what Character.AI, Fireworks, Baseten, and Together AI actually run, sourced and cited.
Explore
Place it
Once you know the silicon, see every product family that fits it and which AWS, GCP, and Azure instances actually offer it.
Explore
See where each silicon option deploys across AWS, GCP, and Azure