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

# Model Distillation API reference

> Choose between the management API and the inference proxy.

Model Distillation exposes two public HTTP surfaces. Both authenticate with a Forge API key, but they have different responsibilities and tenancy controls.

| Surface | Base URL | Use it for |
| - | - | - |
| Management API | `https://forge.coreweave.com/api/distillation/v1` | Providers, projects, versions, routing, datasets, relabeling, fine-tunes, evaluations, and analytics |
| Inference proxy | `https://proxy.training.coreweave.com/v1` | OpenAI-compatible project and direct-model Chat Completions |

## Management API

The UI uses this same API. Resources are durable and asynchronous operations usually return `202 Accepted` with a queued, building, running, or pending state. Poll the resource's GET endpoint rather than holding the request open.

The optional `Forge-Entity` header selects an accessible team. Resource paths and bodies don't repeat entity ownership.

## Inference proxy

Use an OpenAI SDK with the proxy base URL. The `OpenAI-Project: <entity>/<project>` header selects the team and project, and the OpenAI SDKs send it from their `project` option. With the header, `model` set to `"router"` uses the project's weighted routing, and a `{provider}/{model}` value overrides the model within the project. `metadata["forge.project_version"]` selects a project version, and omitting it selects version 1.

## Interactive endpoint pages

The endpoint pages in this section are generated from the checked-in OpenAPI 3.1 specifications. They include request builders, schemas, response codes, and copyable examples for both surfaces.

> [card] [Management API](https://docs.coreweave.com/model-distillation/reference/management.md)
>
> Manage providers, projects, datasets, fine-tunes, evaluations, routing, and analytics.

> [card] [Inference proxy](https://docs.coreweave.com/model-distillation/reference/proxy.md)
>
> Send OpenAI-compatible Chat Completions requests through projects or registered providers.

Agents can start with [`llms.txt`](https://docs.coreweave.com/llms.txt) and use the [Management API reference](https://docs.coreweave.com/model-distillation/reference/management.md) or [Inference proxy API reference](https://docs.coreweave.com/model-distillation/reference/proxy.md) for endpoints, schemas, and examples.

> [!WARNING]
>
> The playground sends real requests. Creating datasets, fine-tunes, and evaluations can consume inference or training resources; routing writes can change production traffic.
