PNYXMIND

Connections

PnyxMind works from the first minute with its own models: Claude, ChatGPT, Gemini, Kimi and DeepSeek. If you like, you can bring your own too. This page shows how, step by step.

If you get stuck, press Help in the app and the PnyxMind assistant guides you live. It will never ask you to type a key into the chat; keys go only into the form field. · Ελληνικά

The services you can connect

You connect inside the app, under Connections. Press "Connect" and it takes you there (if you are not signed in, it asks for your email first).

PnyxMind inside your apps

Models from Hugging Face

Your own key

A model on your computer

Step by step, with pictures

Choose what you want to connect. Each step is one move: the gold frame shows where to press. Your screens may look a little different on your phone or computer.

Connect Claude

  1. Step 1Open Claude. Press your name at the bottom left, then Settings.

  2. Step 2Press Connectors.

  3. Step 3Press Add custom connector. Type PnyxMind as the name and paste https://app.pnyxmind.com/mcp as the address. Press Add.

  4. Step 4Press Connect. PnyxMind opens: sign in with your email or Google and press Allow. Done!

Connect ChatGPT

  1. Step 1Open ChatGPT. Press your name at the bottom left, then Settings.

  2. Step 2Press Apps & Connectors, then Advanced settings, and switch on Developer mode.

  3. Step 3Press Create. Type PnyxMind as the name, paste https://app.pnyxmind.com/mcp as the address and press Create.

  4. Step 4PnyxMind opens: sign in with your email or Google and press Allow.

  5. Step 5In a new chat press +, switch on PnyxMind and ask anything about your cases.

Add free models from Hugging Face

  1. Step 1Sign in to PnyxMind and press Connections on the home screen.

  2. Step 2Under Members from Hugging Face, press Connect with Hugging Face.

  3. Step 3Sign in to Hugging Face and press Authorize. No account? You make one there, free.

  4. Step 4Pick up to 6 models (the "Recommended" ones work best) and press Check and add. Done!

What you can connect

ConnectionWho it is forWho pays for the models
PnyxMind's modelsEveryone. Nothing to set up.You, with PnyxMind credits
Your own keyAnyone with an account at OpenAI, Anthropic, Google or another model company.The model company bills you directly. PnyxMind charges nothing for those calls.
Your own model (Ollama, LM Studio)Anyone running a model on their own computer or server.Nobody. It runs on your computer.
Models from Hugging FaceFor anyone who wants open models like Llama as extra members.Hugging Face bills you directly, with a small free allowance every month. PnyxMind charges nothing.
Your Claude, ChatGPT, Copilot or Le ChatAnyone who wants to join a case from the AI app they already use.The subscription you already have.

1. Your own key

A key (API key) is a long code the model company gives you. With it, a member of your case speaks through your own account.

Steps
  1. In PnyxMind go to Connections → Your own key and pick the company.
  2. Press Get a key. The right page of that company opens; sign in there with your account.
  3. Press Create key, name it e.g. PnyxMind, and copy it.
  4. Back in PnyxMind, paste it into the field and press Check. In a few seconds you see ✓ Works.
  5. In the case, pick for each member whether it runs on PnyxMind credits or on your key.

Where to get a key:

If the check says the key was not accepted, your account with that company usually has no balance or card yet; add one there first. A ChatGPT Plus or Claude Pro subscription is not a key: keys come from the developer pages above.

2. Your own model (Ollama or LM Studio)

If you run a model on your computer, PnyxMind can talk to it through a private https channel. The small PnyxMind Connect program opens the channel, and it closes as soon as you close its window.

Steps with Ollama
  1. Download Ollama from ollama.com/download and open it.
  2. Get a model, e.g. in a terminal: ollama pull llama3.1:8b. Larger models give better answers if your computer can run them.
  3. Download PnyxMind Connect: Windows · Mac and Linux.
  4. Run it. On Windows: right-click → Run with PowerShell. On Mac and Linux, in a terminal: bash ~/Downloads/pnyxmind-connect.sh.
  5. It finds Ollama on its own and prints an address such as https://something.trycloudflare.com (on Windows it also copies it).
  6. In PnyxMind: Connections → Your own model, paste the address, pick the model from the list and press Check.
  7. Keep the window open while the case runs.
Steps with LM Studio
  1. Open LM Studio and load a model.
  2. Go to Developer and switch on Start server.
  3. From there, the same steps as Ollama from step 3.

The address is random and works only while the window is open. PnyxMind takes only public https addresses and checks them again before every call. Your own model can be a member on any case, but never the chair. PnyxMind cannot see which model really runs on your computer, so choosing it is your responsibility.

3. Your Claude, ChatGPT, Copilot or Le Chat

The official PnyxMind connector: from the AI app you already use, open a new case, see your cases, read the minutes and vote, with no keys. The app always asks you before it opens a case, approves or votes for you, and shows you the cost first.

The connector address

https://app.pnyxmind.com/mcp

  1. Claude: Settings → Connectors → Add custom connector. Name it PnyxMind, paste the address, Add.
  2. ChatGPT: Settings → Apps & Connectors → Advanced → turn on Developer mode, then Create and paste the address.
  3. Le Chat: Connectors → Add connector → Custom MCP, paste the address.
  4. Copilot: in your company's Copilot Studio, add a Model Context Protocol tool with the address. If you don't manage it, give this page to your IT team.
  5. The PnyxMind page opens. Sign in as usual, with an email code or Google, and press Allow.
  6. Done. Ask for example "Which of my PnyxMind cases are waiting for me?".

You see which apps have access in PnyxMind, under Connected apps. One press on Disconnect stops the access at once.

coming soon We are applying for PnyxMind to be listed in the official directories of Claude, ChatGPT, Copilot and Le Chat; then you will find it there and just press Connect. On national-security cases, and in organisations that keep them off, the Chinese apps (Kimi, DeepSeek, Qwen) do not see the case, just as with PnyxMind's own Chinese models.

4. Models from Hugging Face

Bring open models like Llama or gpt-oss into your sessions as extra members. PnyxMind charges nothing for them. Hugging Face bills you directly, usually very little, and gives you a small free allowance every month.

  1. In PnyxMind, on the home screen, press Connections and open Members from Hugging Face.
  2. Press Connect with Hugging Face, sign in and press Authorize. No account yet? You can make one there for free.
  3. Pick up to 6 models. The ones marked "Recommended" work well in sessions.
  4. Press Check and add. We test each model with a short question before we keep it.

Hugging Face models always sit as members, never as the chair. Models from Chinese labs (Qwen, DeepSeek, Kimi, GLM) stay out of national-security and defence cases.

Security