Simple & cost-effective batch processing

Batch processing simplified, at the lowest cost

Ship jobs in a declarative JSON/YAML format.
Solti routes each workload to the most efficient provider in real time.

Batch processing simplified

Define jobs declaratively and keep infra out of the way.

Define your compute, volumes, and environment in a familiar YAML or JSON format. No complex infrastructure management or cluster provisioning required. Just hit the API and let us handle the scheduling.

bash + yaml

curl -X POST https://my-project.api.solti.run/jobs \
  -H "X-Solti-API-Key: <api-key>" \
  -H "Content-Type: application/yaml" \
  --data-binary @- << 'EOF'

image: pytorch/pytorch:2.2.1-cuda12.1-cudnn8-runtime
command: ["python3"]
args: ["/mnt/r2/scripts/train.py", "--epochs", "10", "--batch-size", "64"]

env:
  TORCH_OMP_NUM_THREADS: "8"
  PYTHONUNBUFFERED: "1"

volumes:
  r2-storage:
    connection: my-r2-connection
    bucket: my-ml-datasets
    mountPath: /mnt/r2

resources:
  cpu: 8
  memory: 32Gi
  nvidia.com/gpu: 1

EOF

At the lowest cost

Route each workload to the cheapest viable provider.

Stop overpaying for idle GPU and CPU resources. Our routing layer matches each workload to the most cost-effective provider, region, and instance type in real time.

Routing Engine

AWS

GCP

Azure

Be the first to know

Join the waitlist

Join the waitlist and be first in line for early access to Solti.
We will notify you when we launch and provide early access to our platform.