Sign up and create a flock¶
Create a Flockadoo account, deploy a provider flock, and copy the provider API key. You need that key for Installation.
Sign up¶
- Open www.flockadoo.ai.
- Click Sign-up.
- Enter your email, or continue with Google or GitHub.
- Open the Complete your signup email and choose a password (at least 8 characters). You land in the dashboard.
Choose Provider¶
On first login you see Welcome Onboard! and Please, choose your role:
- Provider — earn by offering GPU inference
- Consumer — call models as a buyer
- I want both — if you will do both
Choose Provider (or I want both). You can switch between Provider and Consumer later in the dashboard sidebar.
Deploy a flock¶
- Open My Flocks under Inference.
- Click Deploy a Flock.
The wizard has three steps. The first step is titled Check Compatibility.
1. Check Compatibility¶
Confirm your machine matches Software and hardware requirements.
Then set Flock type:
- Solo Flock — one machine.
- Team Flock — this key is the master. Share the flock page so workers can request an invite. When the slots are filled, start the master first. Workers then have 60 minutes to join. See Team Flocks.
If you pick Team Flock, those machines share a private network. Only create or join one with people you intend to share GPUs with. Read Team Flock networking before you generate a key or accept a join.
2. Software¶
The engine depends on your operating system. You do not install it in this step.
- Linux: vLLM. You need an NVIDIA GPU.
- macOS: vLLM Metal. Apple Silicon only. Intel Macs are not supported. The client setup wizard on a Mac also offers llama.cpp. See Setup: macOS.
- Windows: llama.cpp. It does not serve every catalog model. Encoder and embedding models such as
baai/bge-m3stay off Windows.
Skip any vendor install links. The client setup wizard installs the engine after you run the install command.
3. Client (flock name and API key)¶
- Set Flock name. For example,
flockadoo-flock1. - Set API key label (defaults to
{flock name} API key). - Click generate API key.
- Pick Linux, macOS, or Windows for the install command.
The full key is shown once. Copy it before you leave the page. Store it somewhere safe; you will paste it into the install command next.
The wizard shows an install command with your key filled in. You can copy that, or follow Installation and substitute the key yourself.
You set model prices after the client is running, on the flock Models tab — see Model pricing. Native Windows uses llama.cpp, not vLLM.
Copy the key later from Settings¶
If you already have a flock and need the key again:
- Open My Flocks and select the flock (for example flockadoo-flock1).
- Open the Settings tab.
You can Create API Key if the flock has none, or Re-Generate Key for an existing key. Re-generating immediately invalidates the old secret — update the client config before you expect inference to work.
The full secret is only shown when you create or re-generate it. Settings otherwise shows a masked suffix.
More on keys, pause, and delete: Deleting and managing flocks. To add workers and split earnings, see Inviting nodes and Flock earnings.
Next¶
Install the client with that key: Installation.