Moondream logo

Give robots the vision to see, understand, and act in the real world

Moondream's vision model enables robots to locate objects, navigate environments, and complete tasks autonomously.

Our VLM is low latency and efficient, making it perfect for when you need reasoning 24/7. You could run approximately 3,370 vision queries for just $1 with our cloud API or run for free with Moondream Station.

Perfect for humanoids in the home, robotic arms, pick and place systems, drones, as well as manufacturing, warehouse automation, agriculture, and service applications.

Detect objects that need to be picked up

Robot detects and localizes items in the environment to understand what needs attention and where to act.

Try in playground

Detection target

dirty laundry

Detection output

1825ms • 753 tokens • $0.000270
[
  {
    "x_min": 0.3959978563299161,
    "y_min": 0.6305546994754534,
    "x_max": 0.5209131896133878,
    "y_max": 0.7418793181198544
  },
  {
    "x_min": 0.5072101051686715,
    "y_min": 0.8161747350148886,
    "x_max": 0.7264164832966267,
    "y_max": 0.9902768778883372
  },
  {
    "x_min": 0.024318775288883,
    "y_min": 0.7399470321459622,
    "x_max": 0.2044202276436683,
    "y_max": 0.8631810225949956
  }
]

Code Example

import moondream as md
from PIL import Image

# Initialize with API key
model = md.vl(api_key="your-api-key")

# Load robot camera image
image = Image.open("path/to/robot_camera.jpg")

# Detect objects in the scene
result = model.detect(image, "dirty laundry")
objects = result["objects"]
request_id = result["request_id"]
print(f"Found {len(objects)} objects")
print(f"Request ID: {request_id}")
Target where to place items

Robot identifies the location of tools and containers needed to complete the task.

Try in playground

Detection target

laundry basket

Detection output

670ms • 743 tokens • $0.000245
[
  {
    "x_min": 0.19150848668453413,
    "y_min": 0.6840621086640302,
    "x_max": 0.35981116140930747,
    "y_max": 0.9092907750114341
  }
]

Code Example

import moondream as md
from PIL import Image

# Initialize with API key
model = md.vl(api_key="your-api-key")

# Load robot camera image
image = Image.open("path/to/robot_camera.jpg")

# Detect objects in the scene
result = model.detect(image, "laundry basket")
objects = result["objects"]
request_id = result["request_id"]
print(f"Found {len(objects)} objects")
print(f"Request ID: {request_id}")
Work Area Assessment

Robot detects all items requiring processing in the work area.

Try in playground

Detection target

dirty dishes

Detection output

671ms • 742 tokens • $0.000245
[
  {
    "x_min": 0.4222182512910134,
    "y_min": 0.5794431174211956,
    "x_max": 0.7038814554538547,
    "y_max": 0.9063828845338386
  }
]

Code Example

import moondream as md
from PIL import Image

# Initialize with API key
model = md.vl(api_key="your-api-key")

# Load robot camera image
image = Image.open("path/to/robot_camera.jpg")

# Detect objects in the scene
result = model.detect(image, "dirty dishes")
objects = result["objects"]
request_id = result["request_id"]
print(f"Found {len(objects)} objects")
print(f"Request ID: {request_id}")
Tool Localization

Robot locates cleaning tools needed to execute the task.

Try in playground

Detection target

sponge

Detection output

688ms • 741 tokens • $0.000244
[
  {
    "x_min": 0.6434533456694933,
    "y_min": 0.5320376018806122,
    "x_max": 0.7426659114175058,
    "y_max": 0.6272976864869343
  }
]

Code Example

import moondream as md
from PIL import Image

# Initialize with API key
model = md.vl(api_key="your-api-key")

# Load robot camera image
image = Image.open("path/to/robot_camera.jpg")

# Detect objects in the scene
result = model.detect(image, "sponge")
objects = result["objects"]
request_id = result["request_id"]
print(f"Found {len(objects)} objects")
print(f"Request ID: {request_id}")
Get Started

Get Running in Minutes.

Moondream is open source and you can install and run it anywhere, for free. You can have it running on your computer or in our cloud in a matter of minutes.

Run It Yourself
  • Moondream Station is free
  • Works with our Python and Node clients
  • Works offline, fully under your control
  • CPU or GPU compatible
Moondream Station
Run in the Cloud
  • Spin up instantly—no downloads or DevOps
  • $5 in free monthly credits, no card required
  • Predictable pay-as-you-go pricing
  • 2 RPS on free tier, scales to 10 RPS or more with paid credits
Moondream Cloud