Provision compute instance
Returns x402 and MPP 402 payment challenges when no payment credential is supplied, or provisions a Vultr or DigitalOcean instance after x402 or MPP payment settlement.
Body
Prepaid runtime in hours (minimum 24). Provision uses prepaid_hours; the extend endpoint uses extend_hours.
x >= 24Pay from prepaid USD credits instead of an x402 wallet payment. Supported for bare and grid machines. NOT supported for ai_machine (private AI Machines require an x402 wallet payment).
Optional provider hint. Plan IDs with do: prefix resolve to DigitalOcean.
vultr, digitalocean x402 payment network. Robinhood Chain settles in USDG; Base and MegaETH settle in USDC; Solana settles in USDC. MPP clients use Authorization: Payment and do not need this field.
base, solana, megaeth, robinhood Optional cloud-init user data run at first boot.
Private AI Machine (Standard tier). When present, the machine self-installs the pinned LLM behind your own OpenAI-compatible endpoint (URL + API key), running llama-server (llama.cpp) on port 8080 exposing /v1 (/v1/chat/completions and /v1/models). Supports x402, MPP, or prepaid credits. Mutually exclusive with deploy_node.
Join the grid. When present, the machine self-installs as a network node owned by the paying wallet, earning USDC + SGL (requires 50,000 SGL staked). Requires an x402 Solana payment. Mutually exclusive with ai_machine.
Response
Provisioning started
