BibGenie DocsUsing Ollama in Zotero
Zotero Ollama Setup Guide - Configure DeepSeek, Qwen, Llama local AI models in Zotero for free, private AI-assisted research
Ollama is the easiest way to run local large language models (such as DeepSeek, Qwen, Llama, Mistral, etc.). By configuring Ollama models in Zotero, you can run AI models locally without relying on cloud services, protecting your privacy while saving costs.
Prerequisites
Before you begin, make sure you have installed the following software:
- Zotero - Reference management software
- Ollama - Local model runtime tool
- BibGenie Plugin - Zotero AI assistant plugin
Verify Ollama Installation
First, verify that Ollama is installed correctly. Open your terminal and enter the following command:
ollama listIf the installation is successful, you will see output similar to:
NAME ID SIZE MODIFIED
qwen2.5:3b 357c53fb659c 1.9 GB 8 months ago
llama3:latest a6990ed6be41 4.7 GB 1 months ago
phi3:latest a2c89ceaed85 2.3 GB 1 months agoThis output indicates that Ollama has installed three models: qwen2.5:3b, llama3:latest, and phi3:latest.
Installation Failed?
If the command fails or shows that ollama is not found, please refer to the Ollama Official Installation Guide for installation instructions.
Start Ollama Service
After verifying the installation, you need to start the Ollama service. Enter the following command in your terminal:
ollama serveWhen the service starts successfully, you will see output similar to:
time=2025-12-07T14:06:04.605+08:00 level=INFO source=routes.go:1331 msg="server config" ...
time=2025-12-07T14:06:04.643+08:00 level=routes.go:1384 msg="Listening on 127.0.0.1:11434 (version 0.11.8)"Default Port
The Ollama service listens on 127.0.0.1:11434 by default. If you need to change the port or allow remote access, please refer to the Ollama official documentation.
Configure Ollama in BibGenie
Once the Ollama service is running, you can configure the model in Zotero.
Open Model Settings
Click the gear icon in the top right corner of the BibGenie plugin window, then select "llm-settings" to open the model settings page.
Choose Configuration Method
You can configure Ollama models in two ways:
- Create New Model: Click the Add Model button in the top left to create a new model
- Edit Built-in Model: Find the built-in Ollama model in the Custom Models list and edit it directly
Fill in Configuration
Based on your chosen configuration method, fill in the appropriate settings.
Save and Enable
After configuration, click save. If the model is disabled, remember to turn on the model switch.
Configuration Options
When creating a new Ollama model, configure the following options:
| Option | Required | Description |
|---|---|---|
| Model Name | ✓ | Display name for the model, e.g., Ollama-qwen2.5:3b |
| Model ID | ✓ | Ollama model ID, the value from the NAME column in ollama list |
| Provider Name | ✓ | Provider name, e.g., Ollama |
| Provider Type | ✓ | Select Ollama |
| Base URL | - | API address, auto-filled as http://127.0.0.1:11434 when Ollama is selected |
| API Key | - | API key, empty by default unless you specified one at startup |
| Model Type | ✓ | Model type, usually select text, if the model supports other features, you can select other features |
| Max Tokens | - | Maximum token count, adjust based on model, default not filled |
| Context Window | - | Context window size, adjust based on model, default not filled |
| Description | - | Model description, customizable, default not filled |
When editing a built-in Ollama model in BibGenie, you only need to configure:
| Option | Required | Description |
|---|---|---|
| Model Name | ✓ | Display name for the model |
| Model ID | ✓ | Enter the Ollama model ID you want to use |
| API Key | - | API key, empty by default |
| Base URL | - | API address, defaults to http://127.0.0.1:11434 |
Start Using
After configuration, you can use the configured Ollama model in BibGenie's chat interface:
- Open the BibGenie chat window
- Select your configured Ollama model from the model selection dropdown below the input box
- Start chatting with your local AI model
Recommended Models
For research literature reading and translation tasks, we recommend the following models:
- qwen3 - Balanced performance and speed
- llama3 - Strong English comprehension
- deepseek-r1 - Excellent reasoning capabilities