> This is the markdown version of https://www.maniac.ai/docs. Visit the full page for interactive content.


[

![](/layout/logo-icon.svg)![](/layout/logo-dark.svg)![](/layout/logo-wordmark-light.svg)![](/layout/logo-wordmark-dark.svg)

](/)

Search

⌘K

[Introduction](/docs)[Quickstart](/docs/overview/quickstart)[Installation](/docs/overview/installation)[Authentication](/docs/overview/authentication)

[Agent Containers](/docs/containers)

[Evaluations](/docs/evaluations)

[Judge evals](/docs/evaluations/judge-evals)[Code evals](/docs/evaluations/code-evals)

Optimization

[Create a dataset](/docs/optimization/datasets)[Launch an optimization run](/docs/optimization/runs)[Observability](/docs/optimization/observability)

Routing

Integrations

Guides & Examples

[Guides](/docs/guides/guides)[Tutorials](/docs/guides/tutorials)[Examples](/docs/guides/examples)

[API Reference](/docs/api-reference)

# Introduction

Outperform frontier models on your niche domain tasks at 1/100 the cost. Get started with Maniac in minutes.

Copy MarkdownOpen

* * *

pythontypescript

### [Install the library](#install-the-library)

```
npm install maniac-js
```

### [Initialize the client](#initialize-the-client)

Requires an API key

```
const maniac = new Maniac()
```

### [Run inference](#run-inference)

```
const maniac = new Maniac()

const response = await maniac.chat.completions.create({
  model: "openai/gpt-5",
  messages: [{
      role: "user",
      content: "Hello!"
  }]
})
```

### [Run inference in a container](#run-inference-in-a-container)

```
const maniac = new Maniac()

const container = await maniac.containers.create({
  label: "my-container"
  initial_model: "openai/gpt-5",
})

const response = await maniac.chat.completions.create({
  container,
  messages: [{
      role: "user",
      content: "Hello!"
  }]
})
```

### [That's it! You're all set up with Maniac.](#thats-it-youre-all-set-up-with-maniac)

Monitor your usage and optimization progress in the dashboard.

### [Install the library](#install-the-library-1)

```
pip install maniac
```

### [Initialize the client](#initialize-the-client-1)

Requires an API key

```
maniac = Maniac()
```

### [Run inference](#run-inference-1)

```
maniac = Maniac()

response = maniac.chat.completions.create(
  model = "openai/gpt-5",
  messages = [{
      role: "user",
      content: "Hello!"
  }]
)
```

### [Run inference in a container](#run-inference-in-a-container-1)

```
maniac = Maniac()

container = maniac.containers.create(
  label = "my-container"
  initial_model = "openai/gpt-5",
)

response = maniac.chat.completions.create(
  container = container,
  messages = [{
      role: "user",
      content: "Hello!"
  }]
)
```

### [That's it! Your all set up with Maniac.](#thats-it-your-all-set-up-with-maniac)

Monitor your usage and optimization progress in the dashboard.

[

Quickstart

Set up high throughput, task-specialized agents that outperform frontier models at 1/100 the cost.

](/docs/overview/quickstart)

---

*Maniac, High throughput background agents. Opus-quality outputs at 1/50 of the cost. Learn more at [maniac.ai](https://www.maniac.ai).*