Ultralytics Platform#

Ultralytics Platform is a comprehensive end-to-end computer vision platform that streamlines the entire ML workflow from data preparation to model deployment. Built for teams and individuals who need production-ready computer vision solutions without the infrastructure complexity.

What is Ultralytics Platform?#

Ultralytics Platform brings dataset management and annotation, experiment tracking, cloud and remote training, model export, dedicated inference endpoints, and deployment monitoring into one workspace. It has native support for YOLO26, YOLO11, YOLOv8, and YOLOv5 models.

Workflow: Upload → Annotate → Train → Export → Deploy#

The Platform provides an end-to-end workflow:

graph LR
    subgraph Data["📁 Data"]
        A[Upload]:::start --> B[Annotate]:::proc
        B --> C[Analyze]:::proc
    end
    subgraph Train["🚀 Train"]
        D[Configure]:::proc --> E[Train on GPU]:::proc
        E --> F[View Metrics]:::out
    end
    subgraph Deploy["🌐 Deploy"]
        G[Export]:::proc --> H[Deploy Endpoint]:::proc
        H --> I[Monitor]:::out
    end
    Data --> Train --> Deploy

    classDef start fill:#4CAF50,color:#fff
    classDef proc fill:#2196F3,color:#fff
    classDef out fill:#9C27B0,color:#fff

What you can do:

• Upload images, videos, and dataset files to create training datasets
• Visualize annotations with interactive overlays for all 6 YOLO task types (see supported tasks)
• Train models on cloud GPUs (24 on all plans, 26 with Pro or Enterprise for B200 and B300) with real-time metrics
• Export to 20 deployment formats (ONNX, TensorRT, CoreML, LiteRT, Hailo, Ascend, etc.)
• Deploy to 42 global regions with one-click dedicated endpoints
• Monitor training progress, deployment health, and usage metrics
• Collaborate by making projects and datasets public for the community

Multi-Region Infrastructure#

Your data stays in your region. Ultralytics Platform operates infrastructure in three global regions:

You select your data region during onboarding. Datasets, models, and managed training data remain in that region. Dedicated endpoints are deployed separately to a region you choose from the global deployment map.

Key Features#

Data Preparation#

• Dataset Management: Upload images, videos, or dataset files with automatic processing
• Annotation Editor: Manual annotation for all 6 YOLO task types (detect, segment, semantic, classify, pose, OBB; see supported tasks)
• Skeleton Templates: Built-in (Person, Hand, Face, Dog, Box) and custom skeleton templates for one-click pose annotation
• Smart Annotation: Use SAM 2.1 (Tiny, Small, Base, Large), SAM 3, pretrained Ultralytics YOLO models, or your own fine-tuned YOLO models from the annotation toolbar for detect, segment, semantic, and OBB tasks
• Dataset Versioning: Create numbered NDJSON snapshots with descriptions for reproducible training
• Statistics: Class distribution, location heatmaps, and dimension analysis

graph LR
    A[Upload Dataset/Images/Video]:::start --> B[Auto-Process]:::proc
    B --> C[Browse & Filter]:::proc
    C --> D{Annotate}:::decide
    D --> E[Manual Tools]:::proc
    D --> F[SAM Smart]:::proc
    D --> G[YOLO Auto-Label]:::proc
    E --> H[Train-Ready Dataset]:::out
    F --> H
    G --> H

    classDef start fill:#4CAF50,color:#fff
    classDef proc fill:#2196F3,color:#fff
    classDef decide fill:#FF9800,color:#fff
    classDef out fill:#9C27B0,color:#fff

Model Training#

• Cloud Training: Train on cloud GPUs (24 on all plans, 26 with Pro or Enterprise for B200 and B300) with real-time metrics
• Remote Training: Train anywhere and stream metrics to Ultralytics Platform
• Project Organization: Group related models, compare experiments, track activity
• 20 Export Formats: ONNX, TensorRT, CoreML, LiteRT, Hailo, Ascend, and more (see supported formats)

You can train models either through the web UI (cloud training) or from your own machine (remote training):

Deployment#

• Inference Testing: Test models directly in the browser with custom images
• Dedicated Endpoints: Deploy to 42 global regions with scale-to-zero by default (single active instance)
• Monitoring: Real-time metrics, request logs, and performance dashboards

graph LR
    A[Trained Model]:::start --> B{Action}:::decide
    B --> C[Browser Predict]:::proc
    B --> D[Export Format]:::proc
    B --> E[Deploy Endpoint]:::proc
    D --> F[ONNX / TensorRT / CoreML / LiteRT / ...]:::out
    E --> G[42 Global Regions]:::proc
    G --> H[API Endpoint URL]:::proc
    H --> I[Monitor & Scale]:::out

    classDef start fill:#4CAF50,color:#fff
    classDef proc fill:#2196F3,color:#fff
    classDef decide fill:#FF9800,color:#fff
    classDef out fill:#9C27B0,color:#fff

Once deployed, call your endpoint from any language:

import requests

url = "https://your-endpoint-url/predict"
headers = {"Authorization": "Bearer YOUR_API_KEY"}

with open("image.jpg", "rb") as f:
    response = requests.post(url, headers=headers, files={"file": f})

print(response.json())

Account Management#

• Teams & Organizations: Collaborate with team members, manage roles and invites
• API Keys: Secure key management for remote training and API access
• Credits & Billing: Pay-as-you-go training with transparent pricing
• Activity Feed: Track all account events and actions
• Trash & Restore: 30-day soft delete with item recovery
• GDPR Compliance: Data export and account deletion

Quick Links#

Get started with these resources:

• Quickstart: Create your first project and train a model in minutes
• Datasets: Upload and manage your training data
• Annotation: Label your data with manual and AI-assisted tools
• Projects: Organize your models and experiments
• Cloud Training: Train on cloud GPUs
• Inference: Test your models
• Endpoints: Deploy models to production
• Monitoring: Track deployment performance
• API Keys: Manage API access
• Billing: Credits and payment
• Activity: Track account events
• Trash: Recover deleted items
• REST API: API reference

Troubleshooting#

Dataset Issues#

Training Issues#

Deployment Issues#

Common Questions#

FAQ#