How to Create LLM API Keys

What vfopro uses an LLM for

vfopro’s AI assistant lets you work on your model in plain language — for example “select all beams on level 2”, “assign a section to the selected elements”, or “how many elements are in the current model?”. Behind the chat window is a large language model (LLM) hosted by an AI provider. It turns your request into steps that vfopro carries out in your model.

The assistant is the only part of vfopro that uses an LLM. Modeling, analysis and results viewing all run on your computer and do not need a key, so you can skip this guide until you want the assistant.

Supported providers

For now, vfopro supports Google, Anthropic and OpenAI. Pick the one you already use or prefer — the assistant works the same way with all three. Create your key by following the provider’s own guide:

Create your key with the provider

1. Open your provider’s guide from the table above and sign in, or create an account with that provider.

2. Follow the guide to create an API key. Providers differ on billing: many ask for a payment method or prepaid credits before a key works, and free tiers change over time, so follow the current instructions in the provider’s guide.

3. Copy the key straight away. Some providers, such as Anthropic, show the full key only once when you create it. If you lose it, you can’t view it again — just create a new one.

Add the key to vfopro

Open Settings with the gear icon in the bottom-right corner of the main window, then click the LLM tab. (On a first run, Set up next to AI Assistant key in the welcome checklist opens the same tab.)

1 2 3 4 5

1. Under Provider, choose OpenAI (ChatGPT), Google Gemini or Anthropic Claude.

2. Paste your key into API Key. Click Show if you want to check what you pasted.

3. Choose a Model. Unless you have a reason not to, pick the one marked (Recommended). The list changes with the provider you selected and is updated as vfopro is updated.

4. Click Test Connection. vfopro sends a small test request to the provider using the key.

   You should see: “Connection successful!” under the buttons.

5. Click Save at the bottom of Settings.

   You should see: “API key is saved (encrypted)” the next time you open this tab.

Where your key is stored

vfopro stores your key encrypted on your computer, using your operating system’s protection, and only uses it to make requests to the provider you chose. It never saves a key unencrypted — if encryption isn’t available on your system, saving fails with an error instead.

To remove a key, select the provider and click Clear API Key. That deletes vfopro’s copy. To make the key stop working everywhere, also delete or revoke it in your provider’s console.

What the assistant sends to the provider

Each time you send a message, vfopro sends the provider:

• your message and the recent conversation, plus a short running summary of earlier turns;
• a compact description of your model — the names of materials, sections, stories and load patterns, element counts by type, the model’s overall extents, and a few sample nodes and elements;
• what you currently have selected;
• any images, text files or markdown files you attach to the message;
• the output of commands the assistant runs for you (shortened first), so it can decide the next step; and
• vfopro’s own instructions for the assistant, including any skills you activate.

vfopro does not upload your model file or your analysis output files. Values the assistant reads from your model or results while working on a request — a drift value it looks up for you, say — can be part of the output that is sent back. If a project is confidential, read your provider’s data-usage terms first.

Try it

Open the assistant with the AI Assistant (VFO) button — the alien-face icon — in the bottom-right corner, next to the Settings gear.

1 2

Type a first request, such as “How many elements are in the current model?”, and press Enter.

Troubleshooting

Still stuck? Email us with the message you see and which provider you chose — never send your key itself.