Forced to Choose
Ollama? NVIDIA only. ComfyUI? NVIDIA only. You can only use ONE GPU vendor at a time. Wasted hardware.
Vendor lock-inHeterogeneous Hardware Support
Stop VendorLock-In.
Mix and match different GPU types in one unified system. NVIDIA CUDA, Apple Metal, AMD ROCm, and CPU - all working together. No vendor lock-in. Use what you have.
CUDA + Metal + ROCm + CPUNo vendor lock-inUse what you haveFuture-proof
CUDA + Metal + CPUNo vendor lock-in
The Vendor Lock-In Trap
Most AI tools force you into one GPU ecosystem.
Use Everything You Have.
rbee orchestrates all GPU types in one unified system. No vendor lock-in.
How It Works
rbee automatically detects and uses all available backends
Configuration Examples
See how easy it is to mix different GPU types
yaml
# rbee.yaml - Heterogeneous hardware setupworkers: - name: gaming-pc host: gaming-pc.local backend: cuda gpus: ["NVIDIA RTX 4090"] - name: mac-studio host: mac-studio.local backend: metal gpus: ["Apple M2 Ultra"] - name: old-laptop host: old-laptop.local backend: cpu cpus: ["Intel i7"]
# rbee automatically routes to best backendUse Cases
How users leverage heterogeneous hardware with rbee
Stop Vendor Lock-In.
Use all your GPUs together. NVIDIA + Apple + AMD + CPU = one unified system. No vendor lock-in.