Deploying an AI Rig

Provision a private GPU inference node for Qwen3-14B in one click.

Tolka Edge allows you to provision dedicated AI Rigs for model inference. Rather than sharing serverless endpoints with thousands of other users (which leads to unpredictable latency and data privacy risks), you get your own isolated GPU.

Deploy in one click

You do not need to write any code or manage any infrastructure to deploy a Rig.

Navigate to the Dashboard

Click Deploy

Click the Deploy Rig button. You will be prompted to select a model profile. Currently, we offer an optimized profile for Qwen3-14B.

Booting

Your Rig will begin provisioning. A fresh GPU instance is allocated, the vLLM container is pulled, and the model weights are loaded into VRAM. This process typically takes 60–90 seconds.

Ready

Once the status turns to Ready, your dedicated endpoint is live and ready to accept requests!

Sending your first request

Once your Rig is deployed, you address it through the standard OpenAI-compatible completions endpoint.

from openai import OpenAI
 
client = OpenAI(base_url="https://api.tolkaedge.com/v1", api_key="tk-live-...")
 
resp = client.chat.completions.create(
    model="Qwen/Qwen3-14B",
    messages=[{"role": "user", "content": "Hello from a private AI Rig."}],
)
print(resp.choices[0].message.content)

import OpenAI from "openai";
const client = new OpenAI({ baseURL: "https://api.tolkaedge.com/v1", apiKey: "tk-live-..." });
 
const resp = await client.chat.completions.create({
  model: "Qwen/Qwen3-14B",
  messages: [{ role: "user", content: "Hello from a private AI Rig." }],
});
console.log(resp.choices[0].message.content);

curl https://api.tolkaedge.com/v1/chat/completions \
  -H "Authorization: Bearer tk-live-..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "Qwen/Qwen3-14B",
    "messages": [{"role": "user", "content": "Hello from a private AI Rig."}]
  }'

Cold Starts

If you haven't sent a request in a while, your Rig may enter a paused state to save you money. When this happens, you don't need to manually click deploy again. Simply send a completion request as normal.

The API gateway will detect that your Rig is paused, hold your HTTP connection open, trigger a background boot of the instance, and then stream the response back to you once the Rig is healthy. Your application code doesn't need to change at all.