Skip to content

LM Studio mode

DuckTapeKiller edited this page Sep 13, 2026 · 2 revisions

LM Studio mode talks to the local server built into LM Studio. Use it if you already download and manage models in LM Studio: Dive lists them, loads the one you pick, and adds its tools and skills on top.

Before you start

  • LM Studio with at least one model downloaded.
  • LM Studio's local server running. Start it from LM Studio's Developer or Local Server tab. It listens on port 1234 unless you change it.

1. Switch the mode on

LM Studio mode is off at first. Open Settings, go to the MODES tab and tick LM Studio under ENABLED MODES. Its icon appears in the top-left mode switcher. Click it.

2. Connect to LM Studio

In Settings, on the MAIN tab, check LM STUDIO SERVER URL. The default is http://127.0.0.1:1234/v1. A bare address such as 127.0.0.1:1234 works too: Dive adds http:// and /v1.

3. Pick a model

Press Refresh models, then choose one from MODEL. You don't need to load it in LM Studio first: when you send a message, Dive asks LM Studio to load the chosen model if it isn't loaded yet.

"Automatic" uses whatever chat model LM Studio has loaded, or else the first one it has available. Press Refresh models after you download a new model in LM Studio.

4. Start chatting

Close Settings and type a message. The first message after choosing a model can take longer while LM Studio loads it.

5. Try a tool and a skill

Ask something that needs a lookup:

What time is it in Tokyo right now?

The model calls the time_and_date tool and answers from the result. /calc 12 * 44 runs a tool directly.

Most native tools start switched on. shell_command, macos_control and propose_plugin start switched off; turn them on in Settings > TOOLS. browse_act also starts off, but in 5.0.8 Settings has no switch for it or for browse_read. Tools that run code or act on a web page ask for your permission every time.

Type /skill:name and your text to use a skill from ~/dive/skills or ~/.agents/skills. See Create a skill.

Settings for LM Studio mode

On the MAIN tab, in LM Studio mode:

Setting What it does Default
LM STUDIO SERVER URL The address of LM Studio's local server http://127.0.0.1:1234/v1
MODEL The model to use, or "Automatic". Refresh models reloads the list
NATIVE TOOL CALLING (OPENAI SCHEMA) Sends tools as structured function schemas, which models trained for tool use follow best. Off forces the text prompt. If LM Studio refuses the tools, Dive falls back to the text prompt On
AGENT MODE Plan-first prompting and a larger tool budget. Tool calls appear as a step timeline in the chat Off
AGENT TOOL BUDGET Tool calls per request in Agent Mode, from 1 to 50. Without Agent Mode the limit is 6 25
COLOR PALETTE, FONT FAMILY, FONT SIZE How this mode looks
SAMPLING PARAMETERS Sent with every request temperature 0.3, top_p 0.95, top_k 40, min_p 0.05, repeat_penalty 1.1, presence_penalty 0, frequency_penalty 0, max_tokens -1, seed -1

How it works

For developers and agents:

  • LM Studio and llama.cpp share one code path: the window posts to /api/lmstudio/stream, handled by handleLocalModeStream() in routes/chat.js, which speaks the OpenAI chat completions format.
  • Before a request, ensureLocalChatModelLoaded() asks LM Studio whether the model is loaded and, if not, loads it through /api/v1/models/load, falling back to /api/v0/models/load.
  • The model list comes from LM Studio's own /api/v1/models endpoint. For "Automatic", resolveAutomaticLocalModel() picks a loaded model that isn't an embedding model, or else the first model listed.
  • The settings are saved in ~/dive/local-model-settings.json.

Troubleshooting

Symptom What to do
No models appear Check that LM Studio's server is running and that LM STUDIO SERVER URL matches its port, then press Refresh models
The first reply takes a long time LM Studio is loading the model. Later messages are faster
A model you just downloaded is missing Press Refresh models
The model never calls tools Check NATIVE TOOL CALLING. Some models follow the text prompt better, so try switching it off

Read next

Clone this wiki locally