Rename a Model

Rename your trained model to reflect its current purpose, version, or deployment stage. Renaming a model updates only the display name—all configurations, model weights, and the model ID remain unchanged.

✅

Renaming is safe

• Model ID stays the same — API integrations and deployments continue to work
• All data is preserved — Weights, adapters, configurations, and metrics remain intact
• No downtime — Active inference and deployments are unaffected
• Change anytime — You can rename models as often as needed

When to rename a model

Consider renaming your model when:

• Improved clarity — A more descriptive name helps team members understand its purpose
• Version tracking — Distinguish between model versions (e.g., "PCB Defect Detection v2.1")
• Deployment stages — Indicate status (e.g., "Production Model", "Staging", "Testing")
• Use case identification — Clearly describe what the model does (e.g., "Manufacturing QC - Scratch Detection")
• Project organization — Standardizing naming conventions across your team
• Performance updates — Reflect improvements (e.g., "High Accuracy Product Classifier")

Navigate to Models page

1. Navigate to your training project

2. Click the Models tab in the sidebar

3. Locate the model you want to rename

4. Click the three-dot menu (⋮) on the right side of the model card

5. A dropdown menu appears with the following options:

   • View Run — View detailed run information and metrics
   • Export Model — Download model for deployment
   • Create Deployment — Deploy model to production (if configured)
   • Rename Model — Update model name (select this)
   • Edit Model Key — Update checkpoint identifier
   • Delete Model — Permanently remove model

Rename your model

1. Click Rename Model from the dropdown menu

2. A dialog appears with the current model name in an editable field

3. Enter your new model name

   💡

   Naming best practices

   • Be descriptive — Explain what the model does (e.g., "PCB Defect Detection" instead of "Model 1")
   • Include version — Track iterations with version numbers (e.g., "Product Classifier v2.1")
   • Indicate stage — Show deployment status (e.g., "Production - Quality Control", "Staging - Product Detection")
   • Use consistent format — Follow your organization's naming conventions
   • Keep it concise — Aim for 3-8 words for readability
   • Add context — Include base model or task type if helpful (e.g., "Qwen2VL - Manufacturing QC")

4. Click Save or Confirm to apply the new name

The name updates immediately. Your model now appears with the new name throughout the platform.

✅

Name updated successfully

Your model has been renamed. The new name appears in:

• Models list in your training project
• Run details and training history
• Downloaded models and exports
• SDK queries and API responses
• Deployment configurations
• Team member notifications

What changes when you rename

Updated:

• Display name in the platform UI
• Model name in the Models tab
• Name shown in run details
• Name returned by SDK queries
• Name in export and download operations
• Name in deployment configurations

Unchanged:

• Model ID (Run ID) — Used for API calls, downloads, and integrations
• Model weights and adapter files
• Training configuration and hyperparameters
• Training metrics and evaluation results
• Model performance characteristics
• Active deployments and inference endpoints
• Team member access and permissions

🔑

Model ID remains constant

The model ID (also called run ID) never changes, even when you rename the model. This ensures your API integrations, SDK scripts, deployments, and inference workflows continue to work without requiring updates.

You can find your model ID in the Models tab or by viewing the run details.

Naming conventions for teams

Establish consistent naming conventions to help your team stay organized:

Pattern 1: Task-Version

Simple and clear for small teams:

• Defect Detection v1
• Product Classification v2.1
• Quality Control v3

Pattern 2: Stage-Task-Version

Include deployment stage for production tracking:

• PROD - PCB Inspection v2.0
• STAGING - Defect Detection v2.1
• DEV - Product Classifier v1.5

Pattern 3: BaseModel-Task-Version

Useful when training multiple architectures:

• Qwen2VL - Manufacturing QC v1
• Florence2 - Product Detection v2
• Llama3.2 - Visual Inspection v3

Pattern 4: Project-UseCase-Version

For organizations with multiple projects:

• Manufacturing - Scratch Detection v2.1
• Quality Control - Component Check v1.0
• Packaging - Label Verification v3.2

💡

Choose what works for you

There's no one "correct" naming convention. Choose a pattern that makes sense for your team's workflow and stick to it consistently across projects.

Rename models via Vi SDK

You can update model names programmatically using the Vi SDK:

import vi

# Initialize client
client = vi.Client(
    secret_key="your-secret-key",
    organization_id="your-organization-id"
)

# Get the run (models are identified by run ID)
run = client.runs.get(run_id="your-run-id")

# Update the name
run.update(name="New Model Name v2.0")

print(f"✓ Model renamed to: {run.name}")

# Rename all models from a specific project
project_id = "your-project-id"

for run in client.runs:
    if run.project_id == project_id and run.status.phase == "succeeded":
        # Add prefix to indicate production status
        new_name = f"PROD - {run.name}"
        run.update(name=new_name)
        print(f"✓ Renamed: {run.name}")

# Add version numbers to existing models
models_to_version = {
    "run_abc123": "PCB Defect Detection v1.0",
    "run_def456": "PCB Defect Detection v2.0",
    "run_ghi789": "PCB Defect Detection v2.1",
}

for run_id, new_name in models_to_version.items():
    run = client.runs.get(run_id=run_id)
    run.update(name=new_name)
    print(f"✓ {run_id}: {new_name}")

Common questions

Version tracking examples

Here are practical examples of how to track model versions with names:

Initial:     "Product Defect Detection"
After tuning: "Product Defect Detection v1.1"
Major update: "Product Defect Detection v2.0"
Bug fix:      "Product Defect Detection v2.0.1"
Production:   "PROD - Product Defect Detection v2.0.1"

"PCB Inspection - Qwen2VL - Test A"
"PCB Inspection - Qwen2VL - Test B"
"PCB Inspection - Florence2 - Test A"

"PROD - PCB Inspection - Qwen2VL"  (winner)

Development:  "DEV - Quality Control v3.0"
Staging:      "STAGING - Quality Control v3.0"
Production:   "PROD - Quality Control v3.0"

"Manufacturing QC - 2026-01 - v1"
"Manufacturing QC - 2026-02 - v2"
"Manufacturing QC - 2026-03 - v2.1"

Best practices

Next steps

Related resources

• Manage models — Overview of model management operations
• Edit a model key — Update checkpoint identifiers
• Delete a model — Permanently remove models
• Download a model — Export models for deployment
• Manage runs — View training history and metrics
• Train a model — Learn about training workflows
• Evaluate a model — Assess model performance
• Vi SDK Models API — Manage models programmatically
• Vi SDK Runs API — Access training runs via SDK