Set up VoxelMind

Player2 (free daily AI power)

1. 1

   Create a free account at player2.game. If you use the Player2 app, keep it running.

2. 2

   In VoxelMind settings choose Player2 and click Connect with Player2.

3. 3

   With the Player2 app running one click is enough. Otherwise your browser opens and you confirm a code.

4. 4

   On a server, run /vml player2 instead.

Player2 picks the model and sets the daily allowance. VoxelMind has no say in either.

VoxelMind Sparks

1. 1

   In the VoxelMind settings choose VoxelMind and sign in. The game checks your Minecraft account with Mojang as when joining a server: no password, no email.

2. 2

   Your Welcome Sparks arrive, plus a few free Sparks every day.

3. 3

   Check your balance with /vml sparks. The chat line also has a link to the shop with your name filled in.

4. 4

   Packs are also available on this site, for any Minecraft name, for example as a gift.

On a dedicated server

1. 1

   Sign in once on your own PC in singleplayer, then run /vml sparks and click Copy token for a server.

2. 2

   On the server console or as operator run /vml endpoint voxelmind.

3. 3

   Then run /vml key <token> with the token you copied. Treat the token like a password.

OpenRouter (your own account, any model)

1. 1

   Create an OpenRouter account at openrouter.ai and add some credit.

2. 2

   In VoxelMind settings click Sign in with OpenRouter. Your browser opens and OpenRouter creates a key for VoxelMind in your account. You can also paste an existing key.

3. 3

   Click the model button to browse models. Only models that support tool calls are listed, sorted by estimated cost per hour of play.

4. 4

   Run Test connection to check key, model and tool calls with one tiny request.

On a dedicated server use /vml login, open the link, sign in, then run /vml login <code>. Models: /vml models and /vml model <id>.

You pay OpenRouter directly. The key is stored in your game folder and only sent to OpenRouter.

Local model (Ollama or LM Studio)

1. 1

   Install Ollama or LM Studio, download a model that supports tool calls, and start the local server.

2. 2

   In VoxelMind settings switch the AI source to Ollama or LM Studio. On a server use /vml endpoint ollama or /vml endpoint lmstudio. Any OpenAI-compatible URL works too.

3. 3

   Pick the model in the model list, or with /vml model <id>.

4. 4

   Run Test connection (or /vml test).

Every decision reads roughly 8,000 tokens, so a weak PC can feel slow. Small models often struggle with many tools. If answers time out, a bigger model or a faster PC helps.