Quickstart

Quake AI has no single install command. Before the first build, you set up an account, a project, and a way to reach the platform. Get started below, then choose a path.

Get started#

Work through these steps once, in order. Each links to the page that covers it; this list does not repeat that page's content.

1. Create your account. Sign up with a Rumble.com identity (email/password or SSO) and enable two-factor authentication. See Manage your account.
2. Create your organization. The organization is the container your projects live in. See Manage your account.
3. Create your first project and pick a tier. Choose a datacenter region and add a payment method. Recommended for this quickstart: the Developer plan, the entry tier built for trying the platform (1 shared vCPU, 1 GiB RAM). The Developer plan starts with a single shared vCPU; custom plans can include shared vCPUs too, and dedicated vCPUs sell in a minimum of two for guaranteed, full access. See Resource tiers for the full model and the Quake AI pricing page for current figures. You can add vCPUs, RAM, block storage, object storage, or public IPs at any time, and a kept-running deployment can be inexpensive; see Deployments for what a kept workload costs. None of that is needed for this quickstart: start minimal. See Manage cloud projects.
4. Set up how you will work. Pick the option that matches your plan:
   • Console only: nothing to install. Skip to your first build below.
   • SSH into servers: add an SSH key. See Add an SSH key.
   • CLI or IaC: install the OpenStack client and generate app credentials. See Install the OpenStack client and Generate app credentials.
   • AI assistant: connect the MCP server to load the documentation as context for your coding assistant. No account needed to install. See Provision a server with an AI agent.
5. You are set up. Build your first thing below.

Choose your path#

Deploy an app#

Build and run a containerized web application on an instance, reachable from the internet. This is the fastest route to "my app is live" on the platform today.

Deploy a containerized web application (about 30 minutes)

Provision with your AI assistant#

Drive an MCP-connected coding assistant to pull an infrastructure template, review the plan, and apply it. The MCP server installs without an account, so you can connect a tool and provision in one sitting.

Provision a server with an AI agent (about 10 minutes)

Launch your first server#

Walk the platform fundamentals: create an SSH key, open a security group, launch an instance, and serve a page over SSH. This is the IaaS on-ramp the other paths build on.

Launch your first server (about 30 minutes)

Learning paths#

Each learning path is testable on the Developer plan: a single shared-vCPU instance (or object storage) with no floating IP. Work through them in any order.

Where to go next#

• Platform overview: what Quake AI is, what it runs on, and how the services fit together.
• Deployments: larger, kept-running workloads stood up from templates, with cost estimates.
• Access and credentials: CLI setup, app credentials, API tokens, and credential rotation.
• Automation how-tos: infrastructure-as-code patterns, CI integration, and template usage.