### Container Start Command Example

Source: https://docs.grafilab.ai/compute/templates/custom

An example of a custom start command for a container image, specifically configured to launch Jupyter Lab as root.

```bash
bash -c "jupyter lab --ip 0.0.0.0 --port 80 --allow-root"
```

--------------------------------

### Clone and install from Git

Source: https://docs.grafilab.ai/compute/templates/pytorch

Download a repository and install its dependencies.

```bash
git clone https://github.com/<owner>/<repo>.git
cd <repo>
pip install -r requirements.txt
```

--------------------------------

### Install and Verify PyTorch

Source: https://docs.grafilab.ai/compute/templates/code-server

Install project dependencies like PyTorch and verify the installation and CUDA availability.

```bash
pip install torch
python -c "import torch; print(torch.__version__, torch.cuda.is_available())"
```

--------------------------------

### Expose JupyterLab service

Source: https://docs.grafilab.ai/compute/templates/pytorch

Install and start JupyterLab, binding to all interfaces to allow proxy access.

```bash
pip install jupyterlab
jupyter lab --ip 0.0.0.0 --port 8888 --allow-root
```

--------------------------------

### Expose an Application

Source: https://docs.grafilab.ai/compute/templates/code-server

Start an application on a specific port, binding to all interfaces to ensure it is reachable via the proxy.

```bash
python -m http.server 8080 --bind 0.0.0.0
```

--------------------------------

### Install Python packages

Source: https://docs.grafilab.ai/compute/templates/jupyterlab

Use this command within a notebook cell to install additional libraries. Restart the kernel after installation to make the packages importable.

```python
!pip install transformers accelerate
```

--------------------------------

### Install OpenAI TypeScript SDK

Source: https://docs.grafilab.ai/models/integrations/openai-sdk

Install the official OpenAI package for TypeScript.

```bash
npm install openai
```

--------------------------------

### Install OpenAI Python SDK

Source: https://docs.grafilab.ai/models/integrations/openai-sdk

Install the official OpenAI package for Python.

```bash
pip install openai
```

--------------------------------

### Quick start configuration for Windows (PowerShell)

Source: https://docs.grafilab.ai/models/integrations/claude-code

Set the required environment variables for the Claude Code CLI on Windows using PowerShell.

```powershell
$env:ANTHROPIC_BASE_URL = "https://llm.grafilab.ai"
$env:ANTHROPIC_AUTH_TOKEN = "sk-grafilab-..."
$env:ANTHROPIC_MODEL = "grafilab/qwen3.6-flash"

claude
```

--------------------------------

### Start Claude Code Session

Source: https://docs.grafilab.ai/models/integrations/claude-code

Initiate a session and send a prompt to the configured gateway.

```bash
claude "say hi"
```

--------------------------------

### GET /v1/videos

Source: https://docs.grafilab.ai/models

Lists video generation jobs with cursor pagination.

```APIDOC
## GET /v1/videos

### Description
Lists video generation jobs with cursor pagination.

### Method
GET

### Endpoint
/v1/videos

### Parameters
#### Query Parameters
- cursor (string) - Optional - Pagination cursor.

### Response
#### Success Response (200)
- List of video jobs.

```

--------------------------------

### Quick start configuration for macOS / Linux

Source: https://docs.grafilab.ai/models/integrations/claude-code

Set the required environment variables for the Claude Code CLI on macOS or Linux. Ensure the base URL is the bare origin without the /v1/messages path.

```bash
export ANTHROPIC_BASE_URL="https://llm.grafilab.ai"
export ANTHROPIC_AUTH_TOKEN="sk-grafilab-..."
export ANTHROPIC_MODEL="grafilab/qwen3.6-flash"

claude
```

--------------------------------

### Create and poll video job in Python

Source: https://docs.grafilab.ai/models/integrations/openai-sdk

Demonstrates creating a video job, polling its status until completion or failure, and downloading the resulting content.

```python
video = client.videos.create(model="<video-model-id>", prompt="a paper boat crossing a puddle")

while video.status in ("queued", "in_progress"):
    time.sleep(5)
    video = client.videos.retrieve(video.id)

if video.status == "failed":
    raise RuntimeError(video.error)

content = client.videos.download_content(video.id)
content.write_to_file("out.mp4")
```

--------------------------------

### Verify API Endpoint with cURL

Source: https://docs.grafilab.ai/models/integrations/claude-code

Test the endpoint before starting the CLI to ensure the gateway is reachable. A successful response body starts with {"id":"msg_.

```bash
curl -X POST "$ANTHROPIC_BASE_URL/v1/messages" \
  -H "Authorization: Bearer $ANTHROPIC_AUTH_TOKEN" \
  -H "content-type: application/json" \
  -d '{"model": "grafilab/qwen3.6-flash", "max_tokens": 16, "messages": [{"role": "user", "content": "Say hi."}]}'
```

--------------------------------

### Connect to instance via SSH

Source: https://docs.grafilab.ai/compute/instances

Use these commands to connect to your instance. The specific command depends on whether you are using the Grafilab proxy or connecting directly to the node's IP address.

```bash
ssh <your-slug>@<ssh-proxy-host> -p <port>
# or
ssh <username>@<ip-address> -p <port>
```

--------------------------------

### Chat completion JSON response

Source: https://docs.grafilab.ai/models/quickstart

An example of the JSON structure returned by the chat completion API.

```json
{
  "id": "chatcmpl-...",
  "object": "chat.completion",
  "model": "grafilab/qwen3.6-flash",
  "choices": [
    {
      "index": 0,
      "message": { "role": "assistant", "content": "Hello, good to meet you today." },
      "finish_reason": "stop"
    }
  ],
  "usage": { "prompt_tokens": 13, "completion_tokens": 8, "total_tokens": 21 }
}
```

--------------------------------

### Download a model with huggingface-cli

Source: https://docs.grafilab.ai/compute/templates/comfyui

Use this to download a model from Hugging Face into the checkpoints folder. For gated repositories, run huggingface-cli login first and paste a token.

```bash
huggingface-cli download <repo-id> <filename> \
  --local-dir /opt/ComfyUI/models/checkpoints
```

--------------------------------

### Initialize OpenAI Python client

Source: https://docs.grafilab.ai/models/api-keys

Configure the OpenAI Python client to use the Grafilab API key and base URL.

```python

from openai import OpenAI

client = OpenAI(
    api_key=os.environ["GRAFILAB_API_KEY"],
    base_url="https://llm.grafilab.ai/v1",
)
```

--------------------------------

### GET /v1/models

Source: https://docs.grafilab.ai/models

Retrieves the live model catalogue. Each entry includes Grafilab extensions region and tags.

```APIDOC
## GET /v1/models

### Description
Retrieves the live model catalogue. Each entry includes Grafilab extensions region and tags.

### Method
GET

### Endpoint
/v1/models

### Response
#### Success Response (200)
- List of model entries with id, region, and tags.

```

--------------------------------

### GET /v1/videos/{video_id}

Source: https://docs.grafilab.ai/models

Polls a single video job status. Returns queued, in_progress, completed, or failed.

```APIDOC
## GET /v1/videos/{video_id}

### Description
Polls a single video job status. Returns queued, in_progress, completed, or failed.

### Method
GET

### Endpoint
/v1/videos/{video_id}

### Parameters
#### Path Parameters
- video_id (string) - Required - The ID of the video job.

### Response
#### Success Response (200)
- Job status: queued, in_progress, completed, or failed.

```

--------------------------------

### Verify GPU and Clone Repository

Source: https://docs.grafilab.ai/compute/templates/code-server

Use the integrated terminal to verify GPU availability with nvidia-smi and clone your project repository.

```bash
nvidia-smi
git clone https://github.com/<owner>/<repo>.git
```

--------------------------------

### GET /v1/models/{model}

Source: https://docs.grafilab.ai/models

Fetches a single model entry. Model IDs containing slashes work unencoded.

```APIDOC
## GET /v1/models/{model}

### Description
Fetches a single model entry. Model IDs containing slashes work unencoded.

### Method
GET

### Endpoint
/v1/models/{model}

### Parameters
#### Path Parameters
- model (string) - Required - The model ID.

### Response
#### Success Response (200)
- Model entry with id, region, and tags.

```

--------------------------------

### Create and poll video job in TypeScript

Source: https://docs.grafilab.ai/models/integrations/openai-sdk

Demonstrates creating a video job, polling its status until completion or failure, and downloading the resulting content using async/await.

```typescript
let video = await client.videos.create({
  model: "<video-model-id>",
  prompt: "a paper boat crossing a puddle",
});

while (video.status === "queued" || video.status === "in_progress") {
  await new Promise((r) => setTimeout(r, 5000));
  video = await client.videos.retrieve(video.id);
}

if (video.status === "failed") throw new Error(String(video.error));

const content = await client.videos.downloadContent(video.id);
// content is a Response — stream or buffer its body to a file.
```

--------------------------------

### Send chat completion with Python

Source: https://docs.grafilab.ai/models/quickstart

Install the OpenAI library and use it to send a chat completion request to the Grafilab API.

```bash
pip install openai
```

```python
import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["GRAFILAB_API_KEY"],
    base_url="https://llm.grafilab.ai/v1",
)

response = client.chat.completions.create(
    model="grafilab/qwen3.6-flash",
    messages=[{"role": "user", "content": "Say hello in five words."}],
)

print(response.choices[0].message.content)
```

--------------------------------

### Send chat completion with TypeScript

Source: https://docs.grafilab.ai/models/quickstart

Install the OpenAI library and use it to send a chat completion request to the Grafilab API.

```bash
npm install openai
```

```ts
import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.GRAFILAB_API_KEY,
  baseURL: "https://llm.grafilab.ai/v1",
});

const response = await client.chat.completions.create({
  model: "grafilab/qwen3.6-flash",
  messages: [{ role: "user", content: "Say hello in five words." }],
});

console.log(response.choices[0].message.content);
```

--------------------------------

### Download a model with wget over SSH

Source: https://docs.grafilab.ai/compute/templates/comfyui

Use this to download a model file directly into the checkpoints folder when the Manager does not list it. Run from an SSH session on the instance; the file is placed at /opt/ComfyUI/models/checkpoints/model.safetensors.

```bash
cd /opt/ComfyUI/models/checkpoints
wget -O model.safetensors "https://example.com/path/to/model.safetensors"
```

--------------------------------

### GET /v1/videos/{video_id}/content

Source: https://docs.grafilab.ai/models

Redirects to the finished MP4 video. The redirect target requires no key and the result is kept for seven days after completion.

```APIDOC
## GET /v1/videos/{video_id}/content

### Description
Redirects to the finished MP4 video. The redirect target requires no key and the result is kept for seven days after completion.

### Method
GET

### Endpoint
/v1/videos/{video_id}/content

### Parameters
#### Path Parameters
- video_id (string) - Required - The ID of the video job.

### Response
#### Success Response (302)
- Redirect to the MP4 file.

```

--------------------------------

### Download data from the internet

Source: https://docs.grafilab.ai/compute/templates/pytorch

Download public datasets directly to the instance for faster transfer speeds.

```bash
wget https://example.com/dataset.tar.gz
huggingface-cli download <repo-id> --local-dir ./data
```

--------------------------------

### Configure background and subagent models for macOS / Linux

Source: https://docs.grafilab.ai/models/integrations/claude-code

Set additional environment variables on macOS or Linux to ensure background tasks and subagents use a compatible Grafilab model.

```bash
export ANTHROPIC_DEFAULT_HAIKU_MODEL="grafilab/qwen3.6-flash"
export CLAUDE_CODE_SUBAGENT_MODEL="grafilab/qwen3.6-flash"
```