What Is Baseten (docs.baseten.co) and What Can You Do on This Site?

Updated

docs.baseten.co is the documentation site for Baseten, a platform for model inference and training. It is aimed at developers who want to either call hosted models through an API or deploy and train their own models on production GPU infrastructure. If you only need to send inference requests to a supported model, the docs point you to Model APIs first; if you need control over the model, hardware, or serving code, they point you to dedicated deployments.

What the platform covers

Baseten handles the operational layer around models: model containers, GPU capacity across clouds and regions, scaling, and observability. Its inference engines optimize supported model architectures for production traffic. That means you can focus on the model and the application rather than on provisioning and maintaining serving infrastructure.

The main entry points

The documentation is organized around a few distinct paths:

Path What it is for What you provide
Model APIs Calling hosted endpoints for a curated set of models An API key and an OpenAI-compatible client
Dedicated inference Deploying an open-source, fine-tuned, or custom model on dedicated GPUs A configuration file, a Python model class, or a custom Docker server
Training Fine-tuning with Loops or running your own training code with Training Jobs Training code or a fine-tuning setup
Model Library Finding models, including a dedicated speech endpoint for batch or real-time transcription A model selection

Model APIs support the OpenAI Chat Completions API and the Anthropic Messages API in beta, so you can use familiar client SDKs. Choose this path when a supported model fits your application and you want to start with an API call — Baseten manages the model, engine, hardware, and scaling for you.

When to deploy your own model instead

Dedicated inference is the path when you need control. You define the model in one of three ways:

  • config.yaml — for supported architectures, this can define the model, hardware, and inference engine without custom serving code.
  • A Python model class — when you need custom preprocessing, postprocessing, or dependencies.
  • A custom Docker server — when you need full control over server behavior.

You then deploy with the Baseten CLI. Dedicated deployments give you control over the model, hardware, autoscaling settings, and release lifecycle, and they support stable environments for development, staging, and production.

How to get started

The quickstart path is to make your first inference call: call a hosted model through Model APIs with an API key and the OpenAI SDK. From there, the docs branch depending on your goal — call a hosted model, deploy your own model, or train a model. There is also a walkthrough for deploying a Hugging Face model with config.yaml and the Baseten CLI.

If you want to point a coding agent at Model APIs, the docs cover that under Coding agents, and there is separate guidance for letting an agent manage your workspace.

Practical notes

  • The site is documentation, not the product console itself — expect guides, references, and quickstarts rather than a dashboard.
  • Pricing and limits are documented separately, as are billing and usage details, so check those pages before committing to a path.
  • The platform is positioned for production traffic, so the docs assume you are comfortable with API keys, SDKs, and deployment configuration.