What is Nodea?

Nodea turns the computers your organization already owns into one private AI inference fleet. Install the Nodea app on your team’s devices. Each one becomes a node that runs an open AI model on its own hardware. Then call the whole fleet through an OpenAI-compatible API, using the SDKs and tools you already use.

You don’t need new hardware or a GPU cloud contract, and you don’t send your prompts to a third-party AI provider.

The problem we’re solving

Teams that want to use AI usually have two options. Both have trade-offs:

• Hosted AI APIs are easy to start with, but every prompt, document and customer message goes to an outside provider. Costs grow with every token, and you depend on someone else’s models, prices and policies.
• Running models yourself keeps data in-house, but means buying or renting GPU servers, then operating them: provisioning, scaling, model deployment and monitoring.

Meanwhile, most organizations already own a lot of computing power that sits idle most of the day. Modern laptops and workstations have fast GPUs and plenty of memory, and can run capable open models well.

Nodea puts that idle capacity to work. It pools the machines you already have into a fleet you control, and handles the hard parts of running it for you.

How it works

1. Install the app. A teammate installs the Nodea desktop app and signs in. Their device joins your organization’s fleet.
2. Nodea assigns a model. Your Admins choose which open models the organization uses. Nodea gives each device a model that fits its memory, downloads it, and loads it.
3. Send requests. Your applications call https://api.getnodea.com/openai/v1 with an API key, exactly as they would call OpenAI.
4. The fleet answers. Nodea routes each request to a device that’s serving the requested model, and streams the response back.

As devices come and go, and as demand for different models changes, Nodea keeps rebalancing the fleet on its own.

What you get

• Private inference. Models run on your organization’s own devices, and requests are only ever served by your devices. Nodea relays requests and responses between your applications and your devices, and doesn’t store their content.
• No new infrastructure. No servers to buy, rent or operate. Capacity grows as more teammates connect their machines.
• Drop-in compatibility. Works with the official OpenAI SDKs and with any tool that lets you set an OpenAI base URL. See Making requests.
• Open models of your choice. Pick from open models on Hugging Face, and enable or disable them for the whole organization in one place.
• Visibility. The web console shows every device, which model it’s serving, what’s happening live, and token usage over time.
• Team controls. Invite people, give them Admin or Member roles, and manage the organization’s API keys centrally.

Who it’s for

Nodea fits organizations that:

• want to use AI on sensitive or internal data, and don’t want it sent to an outside AI provider
• already have capable devices on their team
• want to build internal tools, assistants or automations on an OpenAI-style API without running GPU servers

Good to know

Note

Nodea runs open models, sized to fit on the devices in your fleet. They’re very capable for many tasks, but they aren’t the largest hosted frontier models. Throughput grows with the number of connected devices: each device generates one response at a time.

• The desktop app currently supports Macs with Apple silicon. See Install the desktop app.
• The API covers Chat Completions and Models, text only. See Differences from OpenAI.
• Only connected devices serve requests. A device that’s asleep, offline or paused is skipped.

Next steps

QuickstartFrom sign-up to your first API response.

OrganizationsHow teams, roles and invitations work.