Scale training jobs on io.net by adding GPUs to your cluster via CLI commands or auto-scaling policies. Go from 1 to 100+ GPUs instantly with no reservations—io.net handles distributed training coordination, network configuration, and GPU allocation automatically.

Manual Scaling

# Add GPUs to existing job
io scale training-job --gpus 16

# Or deploy with specific GPU count
io deploy --image pytorch/pytorch:latest \
  --gpu A100 --count 8 \
  --name distributed-training

Auto-Scaling

# Configure auto-scaling
io autoscale training-job \
  --min-gpus 2 --max-gpus 20 \
  --metric gpu_utilization \
  --threshold 85

# Scales up when GPU util > 85% for 5+ minutes
# Scales down when util < 50% for 10+ minutes

Scale GPUs on io.net with instant provisioning and auto-scaling.