Custom providers

A custom provider connects Slashspace to a model server that isn't in the built-in list: a local model in LM Studio or vLLM, your company's AI gateway, or a smaller hosted service. A private or fine-tuned model can then answer on the same canvas as a commercial one.

Messages to a custom model go straight from your computer to that server. They don't pass through Slashspace's servers or draw from the hosted wallet.

What OpenAI-compatible means

The server accepts requests in the format of OpenAI's Chat Completions API: Slashspace sends each message to your base URL plus /chat/completions, with your API key as a bearer token. Most self-hosted servers and gateways support this. For Ollama or the built-in services, use Ollama or API providers.

Add a provider

Custom providers are included with the trial, the yearly subscription and Lifetime (Plans and billing). If your plan doesn't include them, the page shows a lock and an Upgrade button.

1. Open Settings > Custom Providers and click Add Custom Provider.
2. Fill in:
   • ID: a short identifier such as lm-studio. Use letters, numbers and dashes (a colon breaks it). It can't be changed later.
   • Name: the heading its models get in the model picker.
   • API Key (Optional): leave empty if the server doesn't need one. Keys are encrypted on this device.
   • Base URL: the address including its version path, such as http://localhost:1234/v1 (LM Studio) or http://localhost:8000/v1 (vLLM).
3. Click Add Provider. When "Provider added" appears, close the dialog.

Add models

Slashspace can't fetch a model list from a custom server, so you add each model by ID.

1. Expand the provider's row and click Add Model.
2. Enter the Model ID exactly as the server names it (most servers list IDs at the base URL plus /models).
3. Under Modalities, check Text to Text, then click Add Model.

Slashspace sends a short test message and adds the model only if the server answers; otherwise the server's error appears in the dialog. The model then appears in the model picker under the provider's name.

Edit, turn off or delete

• A provider's switch hides or shows its models in the picker.
• Edit changes the name, key or base URL. Leave the key blank to keep it, or check Remove saved API key. Save with Update Provider.
• The trash icon deletes the provider and its models, permanently.
• Each model has its own switch and trash icon. Click a model to set an Alias or System Prompt.

Limitations

Slashspace can't tell what a custom model supports:

• In Agent mode the model is offered tools. If it can't call tools, replies may fail; use Chat mode instead.
• Connected Image nodes and Scratch pads are sent as images, which a text-only model may reject.
• Only a bearer token is sent. Servers that need other headers won't work.
• Image mode and Orchestrator mode don't use custom models.

Troubleshooting

A failed chat shows the error, usually the one your server sent.

• "Cannot connect to API": nothing answered. Check the server is running on that port and, if remote, accepts network connections.
• Not found: the base URL probably lacks its version path, usually /v1.
• Unauthorized: click Edit and re-enter the key.
• Model not found: the Model ID doesn't match the server's /models list.
• Missing from the picker: turn on both the provider's and the model's switch.