One interface for compute, inference, and everything else an ML engineer needs.

Deposit money into Spin and instantly give your model access to GPUs and inference APIs from SF Compute, Modal, Vast.ai, and OpenRouter. We handle the rest and add no platform fees on top.

Each agent gets its own wallet with a budget and an expiry, so spending is scoped to exactly the work you want to pay for. Give every agent in an eval its own wallet and know the maximum bill before you start. You don’t need accounts with Modal, Vast.ai, SF Compute, or OpenRouter. It just works.

With Spin, spinning up compute is as simple as:

uv tool install spin --index https://spin.inductionlabs.com/cli/simple/
spin wallet create --name experiment --budget 25        # on your machine
export CONAV_BROKER_TOKEN=cnv_agent_…                   # inside the agent
spin compute create --provider vastai --spec @lease.json --request-id gpu-001
spin compute exec RESOURCE_ID 'nvidia-smi'

Giving an agent an OpenRouter key is as simple as:

spin openrouter key create
{
  "key": "cnv_agent_…",
  "base_url": "https://api.spin.inductionlabs.com/v1"
}

Point any OpenAI-compatible client at that base URL and every call is billed to the wallet.

Accounts start with $1 of credit and no card. Read the documentation or the API reference.