Feature Request: Enable Online Search Capability for DeepSeek-R1 Models

[gaochangw]

I highly appreciate the efforts of the LM Studio developers in creating such a powerful and versatile local LLM inference solution. This issue was drafted by a local DeepSeek-R1-Qwen-14B model that runs on LM Studio.

Why This is Important:

The integration of the Online Search capability would make LM Studio even more versatile and competitive for developers and businesses who rely on cutting-edge AI solutions. It would also align with the growing trend of integrating external knowledge sources into LLM-based systems.

Proposed Implementation:

1. Configuration Option: Add an option in LM Studio to enable Online Search when loading DeepSeek-R1 models (e.g., enableOnlineSearch: boolean).
2. API Integration: Provide a seamless way to integrate with DeepSeek’s search capabilities through their API or direct model integration.
3. Documentation Update: Include clear documentation on how to use the new feature, including any required dependencies or setup steps.

Use Case Example:

Imagine developers using LM Studio for applications like chatbots, where access to real-time information is critical. With Online Search enabled:

const model = await client.llm.load("deepseek/r1-series", {
  enableOnlineSearch: true,
});

const response = await model.respond([
  { role: "system", content: "Answer the most recent updates on AI advancements." },
  { role: "user", content: "What are the latest developments in AI for 2024?" },
]);