> ## Documentation Index
> Fetch the complete documentation index at: https://docs.orchestrallabs.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Quickstart: FLUX 3 Action DROID

> Deploy the released FLUX 3 Action DROID checkpoint and predict an action chunk.

This quickstart serves the DROID checkpoint's base BF16 package. It expects three RGB camera observations and an eight-value joint state. Review the [checkpoint license](https://huggingface.co/black-forest-labs/flux-3-action-droid/blob/e3b6c915cdcdfb1d5db08e133b8ca3d29be2d7d4/LICENSE.md) before deployment; it is non-commercial.

> **[PI0.5 YAM](/quickstart)**
>
> Three camera images and a seven-value state.

**1. Install and sign in**

Create an environment and install servo-client:

```bash your computer theme={null}
python3 -m venv .venv
source .venv/bin/activate
python -m pip install servo-client
```

Sign in:

```bash your computer theme={null}
servo login
```

Open the login link in a browser. `servo whoami` confirms the active account.

**2. Enroll the robot**

On the robot computer, enroll it with the name `robot-1`:

```bash robot computer theme={null}
python3 -m venv .venv
source .venv/bin/activate
python -m pip install servo-client
```

Enroll it; the command prints a code and waits for approval:

```bash robot computer theme={null}
servo agent install --name robot-1
```

An organization admin approves the printed code on your computer:

```bash your computer theme={null}
servo robot approve CODE
```

The robot computer reports `Enrolled robot robot-1.` when it is ready.

**3. Inspect the checkpoint**

Inspect the pinned repository to confirm the input names and shapes:

```bash your computer theme={null}
servo model inspect hf://black-forest-labs/flux-3-action-droid --revision e3b6c915cdcdfb1d5db08e133b8ca3d29be2d7d4
```

The released DROID contract has three 640×360 RGB images (`observation.images.wrist_image_left`, `observation.images.exterior_image_1_left`, and `observation.images.exterior_image_2_left`) and an eight-value `observation.state`. Use the report for the tensor layout expected by your checkpoint.

**4. Deploy the checkpoint**

Deploy that same pinned checkpoint to the enrolled robot:

```bash your computer theme={null}
servo model deploy hf://black-forest-labs/flux-3-action-droid@e3b6c915cdcdfb1d5db08e133b8ca3d29be2d7d4 --robot robot-1 --wait
```

The default package uses the released BF16 weights and base sampling recipe. Servo reports the deployment ID and waits until it is ready.

**5. Predict an action chunk**

On the robot computer, load observations from your camera and robot code, then send them to the deployment:

```python predict_flux3.py theme={null}
import numpy as np
import servo

sv = servo.Servo()
robot = sv.robots.get("robot-1")
deployment = sv.deployments.get("DEPLOYMENT_ID")
policy = deployment.policy(robot)
inputs = {
    "observation.images.wrist_image_left": np.load("wrist.npy"),
    "observation.images.exterior_image_1_left": np.load("left_exterior.npy"),
    "observation.images.exterior_image_2_left": np.load("right_exterior.npy"),
    "observation.state": np.load("joint_state.npy"),
}
with sv.session(policy, instruction="pick up the red cup") as session:
    chunk = session.predict(inputs=inputs)
print(chunk.horizon, "steps x", chunk.action_dim, "channels; nothing has moved")
```

The returned chunk contains 32 absolute joint-position actions. Prediction returns actions; your application remains responsible for safety checks and sending commands to the robot.

**6. Stop the deployment**

Stop it when you are done:

```bash your computer theme={null}
servo deployment stop --model flux-3-action-droid
```


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.