Use LM Studio With a Browser Extension: Setup Guide
LM Studio is a desktop app for running local AI models with a graphical interface. If you prefer a GUI over the command line, or if you want to try models from Hugging Face without manual conversion, LM Studio is a good choice. This guide shows how to connect it to PageGrok.
Install LM Studio
Download from lmstudio.ai and run the installer. Available for macOS, Windows, and Linux.
LM Studio handles model downloading, quantization selection, and GPU offloading through its interface — no terminal commands needed.
Download a model
- Open LM Studio
- Use the search bar or model browser to find a model
- Pick a quantization level (Q4_K_M is a good default — balances size and quality)
- Click Download
Popular models for reading tasks:
- Gemma 4 — fast and well-rounded
- Qwen 3.5 — strong multilingual support
- Llama 4 — wide community support
- Mistral — compact and efficient
LM Studio shows estimated RAM usage before you download, so you can check compatibility upfront.
Start the local server
- Go to the Local Server tab (labeled “Developer” in some versions)
- Select the model you downloaded from the dropdown
- Click Start Server
The server starts on http://127.0.0.1:1234 by default. You will see a green status indicator when it is ready.
You can verify it works:
curl http://127.0.0.1:1234/v1/modelsThis should return a JSON list of your loaded model.
Connect PageGrok
PageGrok connects to LM Studio through the OpenAI-compatible API:
- Click the PageGrok icon in your browser
- Open Settings (gear icon)
- Change the provider to OpenAI-compatible
- Set the endpoint to
http://127.0.0.1:1234/v1 - Leave the API key field empty (LM Studio does not require one locally)
- Save settings
PageGrok will now use your LM Studio model for all conversations.
LM Studio vs Ollama
Both work well with PageGrok. Here is when to use which:
| LM Studio | Ollama | |
|---|---|---|
| Interface | GUI — visual model management | CLI — terminal commands |
| Auto-discovery | Manual endpoint config | PageGrok detects automatically |
| Model format | GGUF from Hugging Face | Ollama library (pre-packaged) |
| Server start | Click a button | ollama serve |
| Best for | Users who prefer GUI, want to browse Hugging Face models | Users comfortable with terminal, want simplest PageGrok setup |
You can have both installed. Just run one server at a time, or configure PageGrok to point at whichever is currently active.
Troubleshooting
| Symptom | Cause | Fix |
|---|---|---|
| “Failed to connect” | LM Studio server not started | Open Local Server tab and click Start Server |
| “Model not loaded” | No model selected in server tab | Select a model from the dropdown before starting |
| Empty responses | Model too small for the page content | Try a larger model (8B+) |
| Very slow | Model too large for your RAM | Use a smaller quantization or smaller model |
| CORS error | Rare with LM Studio (it allows all origins by default) | Update LM Studio to the latest version |
Switching between providers
PageGrok remembers your last provider setting. To switch:
- To LM Studio: Settings → Provider: OpenAI-compatible → Endpoint:
http://127.0.0.1:1234/v1 - To Ollama: Settings → Provider: Ollama → Endpoint:
http://127.0.0.1:11434(or just set to default)
The change takes effect immediately — no restart needed.
FAQ
Does PageGrok auto-detect LM Studio like it does Ollama?
Not automatically. LM Studio uses the OpenAI-compatible API format on a different port (1234 vs 11434). In PageGrok settings, select 'OpenAI-compatible' as the provider and point it to http://127.0.0.1:1234/v1.
Can I switch between LM Studio and Ollama?
Yes. In PageGrok settings, change the provider and endpoint. You can switch back and forth depending on which server you have running. Only one needs to be active at a time.
Which models work with LM Studio?
LM Studio supports GGUF-format models. Most popular open models on Hugging Face have GGUF versions. Use the built-in model browser to find and download them directly.
Is LM Studio free?
LM Studio is free for personal use. It provides a graphical interface for downloading, managing, and serving local models without using the command line.
Try it on the page you are reading right now
PageGrok connects your local models to any browser tab. Free, private, no sign-up.
Related guides
- Local AI in the Browser: Run Ollama and Chat With Any Webpage — Connect local models from Ollama, LM Studio, or oMLX to the page you are reading. Setup, model recommendations, real limits, and guides for specific reading tasks.
- Connect Ollama to Your Browser: Setup & CORS Fix — Step-by-step guide to connect Ollama to a Chrome extension. Covers installation, CORS configuration with OLLAMA_ORIGINS, and common connection errors.