Settings
Configure provider connections, workspace sandboxes, and agent inference defaults.
AI Engine & Provider
Select WebGPU in-browser engine or OpenAI-compatible server.
Inference Provider
Choose between local in-browser WebGPU or an external inference server.
OpenAI-compatible Server Connection
Configure endpoint parameters for local or cloud inference.
Base URL
API root endpoint (e.g. /api/llm/v1 for proxy, or http://127.0.0.1:11434/v1).
API Key
Bearer token for cloud authorization (optional for local servers).
Model ID
Active model detected from the connected endpoint.
Connection Status
0 models reported by endpoint.
Idle
Server Diagnostics
Verify endpoint accessibility and reload available model IDs.