Auto
Chooses an online CPU worker at submission. No matching worker means the run cannot be queued.
Distributed compute fabric
Compute follows the experiment.
Next Brain matches each validated model to browser, CPU, GPU or multi-node capacity. Auto selects a compatible registered worker and records the actual execution resource with the run.
Resource family
CPU
General-purpose compute for neural simulations, from small experiments to memory-heavy networks.
Single machine
Auto is the defaultCPU · 8 cores
Configured profile · check live worker below
- CPU
- 8 vCPU
- Memory
- 16 GiB
CPU Small
No worker registered
- CPU
- 2 vCPU
- Memory
- 8 GiB
CPU Medium
No worker registered
- CPU
- 8 vCPU
- Memory
- 32 GiB
CPU Large
No worker registered
- CPU
- 32 vCPU
- Memory
- 128 GiB
Distributed · MPI
Auto when a compatible group is registeredCPU Standard
No worker group registered · homogeneous nodes
- CPU
- 16 vCPU
- Memory
- 64 GiB
CPU High Memory
No worker group registered · homogeneous nodes
- CPU
- 32 vCPU
- Memory
- 128 GiB
Resource family
NVIDIA GPU
No NVIDIA worker is connected yet.
Single machine
Auto is the defaultAuto
Chooses an online NVIDIA GPU worker at submission. No matching worker means the run cannot be queued.
NVIDIA GPU Standard
No worker registered
- CPU
- 8 vCPU
- Memory
- 32 GiB
- GPU
- 1
- GPU memory
- 24 GiB
NVIDIA GPU High Memory
No worker registered
- CPU
- 16 vCPU
- Memory
- 64 GiB
- GPU
- 1
- GPU memory
- 80 GiB
Distributed · MPI
Auto when a compatible group is registeredNVIDIA GPU Standard
No worker group registered · homogeneous nodes
- CPU
- 8 vCPU
- Memory
- 32 GiB
- GPU
- 1
- GPU memory
- 24 GiB
NVIDIA GPU High Memory
No worker group registered · homogeneous nodes
- CPU
- 16 vCPU
- Memory
- 64 GiB
- GPU
- 1
- GPU memory
- 80 GiB
Resource family
Apple GPU
Apple GPU profiles with unified memory for supported model backends.
Single machine
Auto is the defaultAuto
Chooses an online Apple GPU worker at submission. No matching worker means the run cannot be queued.
Apple GPU · 16 GiB
Configured profile · check live worker below
- CPU
- 8 vCPU
- Unified memory
- 16 GiB
- GPU
- 1
Apple GPU Standard
No worker registered
- CPU
- 12 vCPU
- Unified memory
- 24 GiB
- GPU
- 1
Apple GPU High Memory
No worker registered
- CPU
- 16 vCPU
- Unified memory
- 64 GiB
- GPU
- 1
Distributed · MPI
Auto when a compatible group is registeredApple GPU Standard
No worker group registered · homogeneous nodes
- CPU
- 12 vCPU
- Unified memory
- 24 GiB
- GPU
- 1
Apple GPU High Memory
No worker group registered · homogeneous nodes
- CPU
- 16 vCPU
- Unified memory
- 64 GiB
- GPU
- 1