LLM Access
Settings → LLM Access stores profiles for language models: Ollama on your PC, Gemini, Groq, OpenAI, OpenRouter, and other OpenAI-compatible providers.
Addons with permission can send chat requests through these profiles. API keys never leave StreamKit+ and are not visible to addons.
You do not need this section unless an addon (or the built-in test chat) should talk to a model.
Global options
| Setting | What it does |
|---|---|
| Enable LLM Access | When off, addons cannot call the model. The test chat on this page still works. |
| Default profile | Auto (best available enabled profile) or a specific profile |
| Allow addons to choose a profile | When off, addons always use the default selection |
| Free-only mode | Auto-routing skips profiles and models marked as paid |
Save the main settings bar after changing these options or editing profiles.
Profiles
Add, edit, or delete a profile:
- Provider, model, optional base URL, API key
- Creativity (temperature) and max response length
- Enabled / paid flags
Refresh models loads the provider’s model list when supported. Detect Ollama looks for a local Ollama install.
If a provider hits a rate limit, auto mode can fall back to another enabled profile.
Test chat
Open test chat on this page: dialog (multi-turn) or a single request, with a profile picker. Use it to confirm a key and model before enabling an addon that depends on LLM Access.