Documentation

Everything you need to run private AI

We are publishing a full documentation portal. While that ships, here is the essential information to get a TensorPanel server up and serving an OpenAI-compatible endpoint in under 10 minutes.

Quick Start

Install TensorAgent

One-curl-command install on Ubuntu 22.04 / 24.04 with NVIDIA drivers. Supports Docker and bare-metal modes.

OpenAI-Compatible API

Drop-in replacement for /v1/chat/completions, /v1/models, /v1/embeddings. Use the OpenAI SDK with a different base_url.

Security & Compliance

Architecture diagrams, data flow, GDPR/KVKK posture, encryption practices.

Install the agent on your GPU server

Get an install token from the dashboard, then run this on your server:

curl -fsSL https://tensorpanel.io/install/<YOUR_TOKEN> | sudo bash

The installer detects your GPU, installs CUDA + nvidia-container-toolkit if missing, registers the agent with TensorPanel, and starts a systemd service.

Call your model with the OpenAI SDK

Once your model is deployed and a tp- API key is created, point the OpenAI SDK at your TensorPanel endpoint:

from openai import OpenAI

client = OpenAI(
    base_url="https://api.tensorpanel.io/v1",
    api_key="tp-...",
)

resp = client.chat.completions.create(
    model="llama-3-70b-instruct",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(resp.choices[0].message.content)

Need help? Talk to us.

We are still expanding the docs. If you cannot find what you need, the team is one email away — usually responding within a few hours.