No speedup. MacBook Pro 13, M1, 16GB, Ollama, orca-mini.
The plugin allows you to open a context menu on selected text to pick an AI assistant action or open the Action Palette to run a one-time action. The most casual AI assistant for Obsidian.
Action Palette
In the Action Palette, Enter submits the prompt when a / or @ search has no results. Escape closes the search menu; Shift+Enter inserts a newline. Clicking the Creativity chip keeps keyboard selection available with the arrow keys and Enter. When text is selected, a replace icon appears next to the chips: click it or press Alt+R (⌥R on macOS) to replace the selection with the answer instead of inserting it below. The choice is remembered until Obsidian restarts.
In Settings → Actions → Edit, each action can either inherit the global provider (Use global setting, the default) or bind a specific configured AI provider. The Action Palette and context menu display a model badge beside an action only when it overrides the global model.
In the Action Palette, your saved actions appear when the prompt is empty. Use the bottom provider/model badge (or /provider and /model) to override the provider or model for the current run only — without modifying the saved action's configured model in settings.
Saved actions keep their prompt, system prompt, creativity, reasoning and replace behavior; freeform palette choices remain separate. Model choices are local preferences and are not included when sharing a community action. An explicit choice never silently switches to another provider, including for images. Choose a vision-capable model for image actions. A removed provider must be selected again; model availability errors are reported by the provider. RAG continues to use the separately configured embedding provider.
No speedup. MacBook Pro 13, M1, 16GB, Ollama, bakllava.
How to use (Ollama)
1. Install Embedding model:
- For English:
ollama pull nomic-embed-text(fastest) - For other languages:
ollama pull bge-m3(slower, but more accurate)
2. Select Embedding provider in plugin's settings and try to use the largest model with largest context window.
- Continue writing
- Summarize text
- Fix spelling and grammar
- Find action items in text
- General help (just use selected text as a prompt for any purpose)
- New System Prompt to create actions for your needs
You can browse and install community actions directly from the plugin settings.
Automatic Updates: Community actions are automatically updated to the latest version. To stick with a specific version or preserve your changes, simply modify the Prompt or System Prompt. This will disable auto-updates for that action. Note: Changing the Name of the action does not disable auto-updates.
Local GPT reads the manually declared modes of each model from AI Providers. Its provider defaults and saved-action overrides apply only in Local GPT. Actions can inherit the provider setting, select a declared native mode, or choose API default to omit the parameter. API default does not mean thinking is disabled.
In the Action Palette, use /reasoning or click the reasoning badge to choose a mode for one request. The choice resets when reopening the palette or changing provider/model. The palette badge previews the selected model's setting; the status beside the generated response shows the actual model and mode after vision routing. A one-request selection cannot carry into another provider or model. Unsupported or stale values are omitted. Community updates preserve local action preferences, and sharing an action excludes them.
The SDK owns native request serialization. Local GPT has no provider mode table, JSON parameter editor, or invented thinking budget. Visible provider-supplied thinking text retains its existing presentation; mode selection does not create or expose hidden reasoning text. No response-time improvement is promised.
Creativity offers API default, which omits temperature entirely. New or legacy
settings without a Creativity value also omit it. Existing explicit choices are
preserved, including None (numeric zero). Actions inherit an explicitly set
global choice unless they specify a number or choose API default themselves.
The palette's API default also overrides the global choice for that request.
No model name affects sampling, and endpoint incompatibility errors remain visible.
- English
- Chinese
- Dutch
- French
- German
- Italian
- Japanese
- Korean
- Portuguese
- Russian
- Spanish
This plugin is available in the Obsidian community plugin store https://obsidian.md/plugins?id=local-gpt
You can also install this plugin via BRAT: pfrankov/obsidian-local-gpt
You also need to install the AI Providers plugin from the plugin store to configure AI providers: https://obsidian.md/plugins?id=ai-providers
Follow the instructions in AI Providers plugin.
- Open Obsidian Settings
- Go to Hotkeys
- Filter "Local" and you should see "Local GPT: Show context menu"
- Click on
+icon and press hotkey (e.g.⌘ + M) - Filter "Local" again and you should see "Local GPT: Action Palette"
- Click on
+icon and press hotkey (e.g.⌘ + J)
- Colored Tags that colorizes tags in distinguishable colors.
- Obsidian AI Providers is a hub for AI providers.
