Smart CodeInsight Options
Go Up to IDE Options
Tools > Options > IDE > Smart CodeInsight
You can configure the Smart CodeInsight integration in the Smart CodeInsigth entry of the IDE node in the Tools Options window, as seen in the image below.
On this window, you can configure your AI integration with the following fields:
- Enable AI Engine checkbox which enables the entire feature.
- Default AI for Editor and the Default AI for Chat Window combo boxes that allow you to choose the available AI for each UI element.
Besides, in the Plugins section, you can configure each AI engine in its corresponding tab with a different set of configuration parameters. Below are the suggested initial options for each engine:
| Configuration | ChatGPT | Ollama** | Gemini | Claude |
|---|---|---|---|---|
| BaseURL | https://api.openai.com/v1 | Depends on the installation.
In case it is a localhost with the default option, the base URL is http://localhost:11434/ |
https://generativelanguage. |
https://api.anthropic.com/v1 |
| APIKey | API Key from OpenAI | API Key from Google AI Studio | API Key from Anthropic | |
| Model* | gpt-4 | codellama | models/gemini-pro | claude-3-5-sonnet-20240620 |
| Timeout | 300000 | 300000 | 300000 | 300000 |
| Maxtoken | 2048 | 2048 | 1024 | |
| Temperature | 0.1 | 0.9 | ||
| TopP | 1 | |||
| TopK | 1 | |||
| Anthropic Version | 2023-06-01 |
Note: (*) RAD Studio now suggests you a list of available models, retrieved from the AI engine previously configured. Currently, this feature is available for all AI engines except Claude.
Attention: (**) In case you are configuring Ollama, you can use a Docker image to install it. For more information check the Configuring Ollama for Smart CodeInsight page.
Tip: To get an engine to work, you must complete the following steps:
- Configure an engine (one of the engines mentioned above).
- Check the Enabled checkbox for that engine.
- Choose that engine for the editor and chat window.