Features

1. Flexible Storage Types

Storage Type	When to use	How to use	Benefits
New EOS Bucket	Starting fresh with a new model repository	1. Select New EOS Bucket during repository creation. 2. TIR automatically provisions a new bucket. 3. Receive access credentials (access key, secret key). 4. Configure uploads using SDK, CLI, or TIR Instances	Fully managed bucket provisioning; automatic credential generation; integrated with TIR dashboard
Existing EOS Bucket	You already have an EOS bucket and want to reuse it	1. Select Existing EOS Bucket during repository creation. 2. Choose from the list of available EOS buckets in your project. 3. The repository links to the selected bucket. 4. Upload files using standard S3-compatible tools	Reuse existing storage infrastructure; consolidate model storage; no additional bucket provisioning
External EOS Bucket	You want to use a bucket from another E2E account or external S3-compatible storage	1. Select External EOS Bucket during repository creation. 2. Provide bucket name, access key, and secret key. 3. Ensure credentials have read/write access to the bucket. 4. Upload files using provided credentials	Cross-account model sharing; integration with external storage systems; flexible credential management

2. Model Versioning

Managing multiple versions of the same model, A/B testing, rollback scenarios.

How to use:

Organize models using folder structures:

# Upload version 1
mc cp -r model-files/* myrepo/bucket-name/v1/

# Upload version 2
mc cp -r model-files/* myrepo/bucket-name/v2/

# Upload production version
mc cp -r model-files/* myrepo/bucket-name/production/

When linking to Model Endpoints, specify the version path:

• Root: / (uses root of bucket)
• Version 1: /v1
• Production: /production

Benefits:

• Easy version management
• Quick rollback capabilities
• Parallel model testing

3. File Browser

Visual inspection of model files, quick file management

How to use:

1. Navigate to Model Repository dashboard
2. Click File Browser tab
3. Browse files and folders
4. Upload files directly via web interface
5. View file metadata (size, last modified)

Benefits:

• Visual file exploration
• No CLI required for basic operations
• Quick verification of uploads

4. Integration with Model Endpoints

Deploying models for inference

How to use:

1. From Model Repository: Use the Deploy Model option in the Model Repository table. You are navigated to the Model Endpoint flow, where you select the framework (e.g., vLLM, SGLang, NVIDIA Triton) and link the repository to the endpoint.

Benefits:

• Easily deploy your custom model as an inference endpoint

---

5. Encryption

Securing model artifacts at rest.

When to use: You need to store sensitive or regulated model weights and want them encrypted at rest.

How to use:

1. Create a new EOS bucket with Enable Encryption selected during repository creation.
2. Upload model files—they are encrypted at rest automatically.
3. Access models as usual.

tip

Encryption must be enabled at bucket creation. Existing buckets cannot be encrypted retroactively.

---