Model providers
SystemSculpt keeps every provider separate behind the scenes while showing a single picker wherever you need it.
Provider types
Pick the category that fits your setup. SystemSculpt treats each provider separately but gives you one place to manage them all:
How: Enable “SystemSculpt provider”, enter your license key and server URL, then pick a model from the catalog in the chat header.
How: Click “Add custom provider”, choose the adapter, paste the API key, and run “Test connection” before saving.
How: Select the HTTP adapter, enter the local base URL (for example http://localhost:11434), and list the exposed models you want in the picker.
Managing providers
Use the Overview & Setup tab to adjust provider settings. Keep these behaviours in mind while you manage the list:
- •Managed and custom providers appear in the same picker while keeping their credentials separate.
- •Favoriting a model surfaces it to the top of pickers and keeps the ribbon indicator in sync.
- •Advanced mode lets you pin different defaults for chat, title generation, and transcription.
Custom provider checklist
When you add a custom provider, the plugin checks your credentials by fetching model lists and, if needed, sending a small test request. Gather the following before you click “Test connection”.
- •Endpoint URL (HTTPS for cloud providers, HTTP for local hosts).
- •API key or bearer token stored in the SystemSculpt settings manager (never hard-coded).
- •Model identifier accepted by the upstream adapter, such as
gpt-4oorwhisper-large-v3. - •Optional organization/team scope for providers like OpenAI when required.
Switching models mid-chat
Switching models mid-chat updates the conversation instantly. When you confirm a new model:
- •The picker hides embedding-only models so you cannot accidentally select them.
- •Your favorites stay at the top of the list and the chat header updates immediately.
- •The chat stores the new selection and refreshes the token counters right away.