Migrate
Bring your local conversations along and keep local inference: Ollama pairing routes your machine's models into the workspace.
Launching soon: this surface is rolling out
| LM Studio | In idapt | How |
|---|---|---|
| Conversations (JSON files on disk) | Chats via the universal chat-JSON importerLM Studio keeps conversations as JSON on disk; map them to the documented shape (a small transform) and upload. | Automatic |
| Downloaded models (GGUF / MLX) | Local AI via Ollama on the same machineOllama manages its own model store: pull the same open models there and pair the machine. | Manual |
| System prompts and presets | Agents | Manual |
| Local OpenAI-compatible server workflows | The hosted gateway's OpenAI-compatible endpointSame protocol, one base-URL change, plus Anthropic- and OpenRouter-compatible endpoints. | Manual |
LM Studio stores chats as JSON files on disk; the app's settings show the data folder they live in.
Match the documented chat-JSON shape (the help article lists the exact fields), then upload at Settings → Import → "Other apps (JSON)".
Install Ollama, pull the models you used, and pair the computer. idapt routes to it as a free local provider, with a hardware-fit filter showing what the machine can serve.
No. idapt's local lane runs through Ollama, which manages its own model store: pull the same open models there and the workspace discovers them on your paired machine.
You keep local (private, no per-token price) and add what a desktop app does not have: the cloud catalog in the same chat, files that persist, agents, and automatic fallback when a job outgrows the machine.