Skip to main content
A short tour of the objects you’ll work with in Valkyrie and how they relate.

​
Models, deployments, and jobs

Fine-tuning job

A training run that adapts an open base model to your dataset using a managed training pipeline. Produces model outputs (weights/adapters) stored for you.

Deployment

A running, servable model behind an OpenAI-compatible endpoint. It can serve a fine-tuned model or any Hugging Face base model.
A typical lifecycle is dataset → fine-tuning job → deployment → API calls, but you can skip straight to a deployment if you just want to serve an existing model. In the dashboard, fine-tunes are grouped into models. A model is one thing you are teaching, and each training run is a version of it (v1, v2, and so on), whatever examples or base model that run used. A deployment is shown as an endpoint, marked Fine-tuned if it serves one of your versions or Base model if it serves a model as it is, with no fine-tune.

​
GPUs are provisioned on demand

Valkyrie doesn’t keep idle GPUs around. When you launch a job or a deployment, it provisions a GPU from a cloud marketplace, runs your workload, and releases it when it’s no longer needed. Deployments stop after 2 hours without requests unless you keep them always on, and can be brought back on demand, so you pay for compute you actually use.
You never manage GPU machines, drivers, or serving infrastructure yourself. Valkyrie handles provisioning, setup, health monitoring, and teardown.

​
Workspaces

Everything you create lives in a workspace: models, endpoints, datasets, API keys, and the wallet. You start with a personal workspace, and can create or join organization workspaces to work with others, each member with a role such as Owner, Builder, or Viewer. A workspace can have a monthly spend limit. See Workspaces and teams.

​
The wallet

Billing is prepaid and per-use. Each workspace has a wallet you add credit to, and its jobs and deployments draw against it, within optional monthly spend limits. Your balance is computed from the full ledger of credits and debits, so every charge and top-up is auditable. See Wallet & billing.

​
Aliases

An alias is a stable, human-readable name for a deployment. Instead of calling a raw deployment slug, you call:
Aliases let you swap the underlying deployment (for example, after redeploying a new version) without changing the URL your applications use. See Aliases.

​
Ways to use Valkyrie

Dashboard

The web app for people: manage data, models, endpoints, keys, and spend.

AI assistant

A chat on Home that turns your material into a fine-tune, with your OK before it spends.

OpenAI-compatible API

Call your deployed models from any OpenAI-compatible client or SDK.

MCP server

Let an AI agent provision, deploy, and operate models programmatically.

​
Authentication at a glance

API keys are prefixed by purpose, for example, deployment keys begin with vk_dep_ and MCP keys begin with vk_mcp_. See Authentication.