Skip to main content
Close-up of a painted hand plucking lute strings with vivid orange and teal neon highlights. Iwant launches an open-weight model on a cloud GPU in your own account. It provisions the server, waits for the model API, and prints an OpenAI-compatible endpoint, an API key, and SSH access. Iwant currently supports GCP. You need a project with billing enabled, GPU quota, and available capacity. The command-line tool is free; GCP charges for the resources you provision.

Quickstart

Launch gpt-oss-20b, send a request, and remove the deployment.

Launch options and recipes

Choose regions, spot capacity, recipe versions, and teardown settings.
Use the Pricing API when comparing catalog inference prices. Those rates are distinct from the GPU infrastructure bill for a model you host yourself. Source and contribution instructions live in the Iwant repository.
Last modified on October 29, 2026