LM Studio 0.4.x Compatibility Mode Added
You can now use the LM Studio API (https://lmstudio.ai/docs/developer/rest) and Stateful Chats with API Key (https://lmstudio.ai/docs/developer/rest/stateful-chats).
MCP Server Added to Frontend Server
The LLM model you use must support Tools.
MCP Server Features Added
- User memory can be used. You can select whether to enable it in the web settings.
-- *Use with caution as it is stored in plain text on your PC. - The MCP server informs the LLM of the actual date and time.
- The Go server handles some processing for Tool Call usage in text-based LLM models.
*Recommended models for Tool usage are qwen/qwen3-vl-30b and openai/gpt-oss-20b in that order.
Frontend Server Management Window Related
- When using LM Studio's Require Authentication, you must also enter the API key in the frontend server management window.
- User API keys must be entered by clicking on a user in the USER MANAGEMENT section.
-- *The LM Studio API key format contains a colon (:), so iOS continuously attempts to save it in lowercase format.
Chat Screen Related
- When using LM Studio mode, SSE is applied. Model loading, prompt generation, and reasoning progress are displayed.
https://github.com/DINKIssTyle/DINKIssTyle-Local-LLM-Gateway-with-TTS/releases
