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

# Model Concepts

> Core concepts behind model onboarding, governance, readiness, and optimization

This page defines the core objects and lifecycle states used by the Models module.

## Core entities

### Model

A catalog entry that represents a cloud model or a local model artifact with metadata, source details, and capability indicators.

### Source types

* **Cloud**: managed provider model.
* **Hugging Face**: repository-based import.
* **URL**: signed or approved artifact URL.
* **Disk**: local or mounted path registration.

### Capabilities and modality

Models can expose capabilities across text, image, and audio input/output, with endpoint compatibility tracked for downstream usage.

### Verification signals

Readiness is informed by status indicators such as:

* model verification,
* security scan outcomes,
* evaluation availability.

## Lifecycle states

```mermaid theme={null}
stateDiagram-v2
    [*] --> Draft
    Draft --> Ingesting: submit onboarding
    Ingesting --> Verified: checks pass
    Ingesting --> NeedsAttention: checks fail
    NeedsAttention --> Ingesting: retry or fix config
    Verified --> Evaluated: run evaluations/benchmarks
    Evaluated --> Published: approved for usage
    Published --> Archived: deprecate or remove
```

## Metadata conventions

Strong metadata improves governance and discoverability:

* ownership and project context,
* tags and tasks,
* source URI/provider details,
* modality and endpoint compatibility,
* links (website, GitHub, Hugging Face) where applicable.

## Operational concepts

### Recommended infrastructure

Model cards can include recommended cluster context for deployment planning.

### Benchmark history

Performance runs help compare throughput and latency profiles across model and cluster combinations.

### Evaluations

Evaluation outputs capture quality signals and support go/no-go decisions for promotion.

## Governance model

```mermaid theme={null}
graph LR
    A[Model Onboarding] --> B[Security and Integrity Checks]
    B --> C[Evaluation and Benchmark Evidence]
    C --> D[Approval Decision]
    D --> E[Published for Teams]
```

## Related pages

* [Quickstart](/models/quickstart)
* [Actions Reference](/models/reference/model-operations-reference)
* [Troubleshooting](/models/troubleshooting)