# API (/bio/api)

<Callout type="warn" title="Draft">
  Outline only.
</Callout>

## Base URL and authentication [#base-url-and-authentication]

## Submit a job [#submit-a-job]

## Check job status [#check-job-status]

## Download results [#download-results]


# Overview (/bio)

<Callout type="warn" title="Draft">
  This section is a skeleton. Replace the placeholders before publishing.
</Callout>

Bio Pipeline runs biology tools and multi-step pipelines on managed GPU infrastructure.

<Cards>
  <Card title="Quickstart" href="/bio/quickstart" description="Submit your first job." />
  <Card title="Tasks" href="/bio/tasks" description="Start from what you want to achieve." />
  <Card title="Tools" href="/bio/tools" description="Reference for each supported tool." />
  <Card title="Pipelines" href="/bio/pipelines" description="Chain tools into reusable workflows." />
</Cards>


# Pipelines (/bio/pipelines)

<Callout type="warn" title="Draft">
  Outline only.
</Callout>

## Concepts [#concepts]

## Node types [#node-types]

## Pipeline schema [#pipeline-schema]

## End-to-end example [#end-to-end-example]


# Quickstart (/bio/quickstart)

<Callout type="warn" title="Draft">
  Outline only.
</Callout>

<Steps>
<Step>

### Upload input files [#upload-input-files]

</Step>
<Step>

### Submit a job [#submit-a-job]

</Step>
<Step>

### Download the results [#download-the-results]

</Step>
</Steps>


# Tasks (/bio/tasks)

<Callout type="warn" title="Draft">
  Add one page per task in this folder.
</Callout>

Each task page starts from a goal, compares the tools that can reach it, and then recommends one for common situations.

Use this structure for every task page:

1. A comparison table of the candidate tools, with inputs, outputs, and capabilities.
2. Recommendations by situation.
3. Links to the tool pages.


# Tools (/bio/tools)

<Callout type="warn" title="Draft">
  Add one page per tool in this folder.
</Callout>

Each tool page is a reference for one model or program.

Use this structure for every tool page:

1. What the tool does and when to use it.
2. Modes, if the tool has more than one.
3. Required and optional parameters for each mode, listed by their API field names.
4. Outputs and an example job.


# CLI (/cloud/cli)

<Callout type="warn" title="Draft">
  Outline only.
</Callout>

## Install [#install]

```bash tab="macOS" tab-group="os"
# <install command for macOS>
```

```bash tab="Linux"
# <install command for Linux>
```

## Authenticate [#authenticate]

## Command reference [#command-reference]


# Compute (/cloud/compute)

<Callout type="warn" title="Draft">
  Outline only.
</Callout>

## Instance types [#instance-types]

## Images [#images]

## Managing instances [#managing-instances]


# Access control (/cloud/iam)

<Callout type="warn" title="Draft">
  Outline only.
</Callout>

## Projects [#projects]

## Users and roles [#users-and-roles]

## API keys [#api-keys]


# Overview (/cloud)

<Callout type="warn" title="Draft">
  This section is a skeleton. Replace the placeholders before publishing.
</Callout>

Supergate Cloud provides GPU compute, storage, and networking for AI training and inference workloads.

<Cards>
  <Card title="Quickstart" href="/cloud/quickstart" description="Create your first GPU instance." />
  <Card title="Compute" href="/cloud/compute" description="GPU instances and instance types." />
  <Card title="Storage" href="/cloud/storage" description="Block, file, and object storage." />
  <Card title="Networking" href="/cloud/networking" description="Networks, subnets, and public IPs." />
</Cards>


# Networking (/cloud/networking)

<Callout type="warn" title="Draft">
  Outline only.
</Callout>

## Networks and subnets [#networks-and-subnets]

## Security groups [#security-groups]

## Public IP addresses [#public-ip-addresses]


# Quickstart (/cloud/quickstart)

<Callout type="warn" title="Draft">
  The commands below are placeholders. Replace them with the real console steps and CLI syntax.
</Callout>

<Steps>
<Step>

### Sign in and choose a project [#sign-in-and-choose-a-project]

Open the console and select the project that will own the instance.

</Step>
<Step>

### Create an instance [#create-an-instance]

<Tabs groupId="interface" persist items={['Console', 'CLI']}>
  <Tab value="Console">
    Go to **Compute → Instances → Create instance**, choose an instance type and image, then select **Create**.
  </Tab>
  <Tab value="CLI">
    ```bash
    <cli> compute instance create --name my-instance --type <instance-type> --image <image-id>
    ```
  </Tab>
</Tabs>

</Step>
<Step>

### Connect over SSH [#connect-over-ssh]

```bash
ssh <user>@<public-ip>
```

</Step>
</Steps>


# Storage (/cloud/storage)

<Callout type="warn" title="Draft">
  Outline only.
</Callout>

## Block volumes [#block-volumes]

## Shared file systems [#shared-file-systems]

## Object storage [#object-storage]


# Supergate Docs (/)

Pick a product to get started. Each product has its own sidebar tab, and search (<kbd>⌘</kbd> <kbd>K</kbd>) covers all of them.

<Cards>
  <Card title="Cloud" href="/cloud" description="GPU infrastructure: compute, storage, networking, and access control." />
  <Card title="Token Factory" href="/token-factory" description="OpenAI-compatible inference API for open models." />
  <Card title="Bio Pipeline" href="/bio" description="Run biology tasks and pipelines without managing infrastructure." />
</Cards>

## For AI agents [#for-ai-agents]

- [`/llms.txt`](/llms.txt) lists every page.
- [`/llms-full.txt`](/llms-full.txt) contains the full text of every page.
- **Copy Markdown** at the top of each page copies that page as Markdown.


# API reference (/token-factory/api-reference)

<Callout type="warn" title="Draft">
  This page is a placeholder. Generate the reference from the OpenAPI spec with `fumadocs-openapi` once the spec is available.
</Callout>


# Endpoints (/token-factory/endpoints)

<Callout type="warn" title="Draft">
  Outline only. Confirm the billing and scaling details before publishing.
</Callout>

| | Serverless | Dedicated |
| --- | --- | --- |
| Billing | Per token | Per GPU hour |
| Capacity | Shared | Reserved for you |
| Best for | Prototyping and variable traffic | Steady, high-volume traffic and custom models |

## Serverless endpoints [#serverless-endpoints]

## Dedicated endpoints [#dedicated-endpoints]


# Overview (/token-factory)

<Callout type="warn" title="Draft">
  This section is a skeleton. Replace the placeholders before publishing.
</Callout>

Token Factory serves open models through an OpenAI-compatible API. Existing OpenAI SDK code works after you change the base URL and API key.

<Cards>
  <Card title="Quickstart" href="/token-factory/quickstart" description="Send your first request." />
  <Card title="Models" href="/token-factory/models" description="Available models and their limits." />
  <Card title="Endpoints" href="/token-factory/endpoints" description="Serverless and dedicated endpoints." />
  <Card title="Migrate from OpenAI" href="/token-factory/migrate-from-openai" description="Switch an existing integration." />
</Cards>


# Migrate from OpenAI (/token-factory/migrate-from-openai)

<Callout type="warn" title="Draft">
  Outline only.
</Callout>

Token Factory accepts OpenAI-compatible requests. In most integrations, only three values change:

| Setting | OpenAI | Token Factory |
| --- | --- | --- |
| Base URL | `https://api.openai.com/v1` | `<TOKEN_FACTORY_BASE_URL>` |
| API key | `OPENAI_API_KEY` | `TOKEN_FACTORY_API_KEY` |
| Model | `gpt-*` | `<model-id>` |

## Differences to check [#differences-to-check]


# Models (/token-factory/models)

<Callout type="warn" title="Draft">
  Outline only. List each model with its ID, context length, and price per million tokens.
</Callout>

| Model ID | Context length | Input / output price |
| --- | --- | --- |
| `<model-id>` | — | — |


# Quickstart (/token-factory/quickstart)

<Callout type="warn" title="Draft">
  `<TOKEN_FACTORY_BASE_URL>` and `<model-id>` are placeholders.
</Callout>

## 1. Create an API key [#1-create-an-api-key]

Create a key in the console and export it:

```bash
export TOKEN_FACTORY_API_KEY="<your-api-key>"
```

## 2. Send a request [#2-send-a-request]

```python tab="Python" tab-group="lang"
import os
from openai import OpenAI

client = OpenAI(
    base_url="<TOKEN_FACTORY_BASE_URL>",
    api_key=os.environ["TOKEN_FACTORY_API_KEY"],
)

response = client.chat.completions.create(
    model="<model-id>",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(response.choices[0].message.content)
```

```js tab="JavaScript"
import OpenAI from 'openai';

const client = new OpenAI({
  baseURL: '<TOKEN_FACTORY_BASE_URL>',
  apiKey: process.env.TOKEN_FACTORY_API_KEY,
});

const response = await client.chat.completions.create({
  model: '<model-id>',
  messages: [{ role: 'user', content: 'Hello!' }],
});
console.log(response.choices[0].message.content);
```

```bash tab="cURL"
curl <TOKEN_FACTORY_BASE_URL>/chat/completions \
  -H "Authorization: Bearer $TOKEN_FACTORY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "<model-id>",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'
```