Skip to main content

Models

1. Text Conversation Models​

These are commonly known as Large Language Models (LLMs). These models are used in intelligent conversation generation, knowledge base Q&A, one-click application creation, and other scenarios. They represent the primary model category in AI applications. You can configure which model to use in your created applications or in workflow nodes that utilize language models.

  • Currently supported general-purpose models include: ChatGPT, Claude, Gemini, DeepSeek, and Qwen.

  • To use these models: Go to [Agent Configuration] - [Model Settings] to set a default model, which will be used for web conversations and channel integrations. You can also set optional conversation models that users can switch between during debugging, conversations, or when other users interact with your agent in the marketplace. Read Agent Configuration - Model Settings to learn more.


Note

Different models have varying parameters and training methods, which affects their performance capabilities. Generally, models with larger parameter counts have stronger understanding of application settings (prompts) and reasoning abilities, but respond more slowly. For example, in the GPT series, gpt-4.1-nano, gpt-4.1-mini, and gpt-4.1 have progressively increasing parameters, improving reasoning capabilities but with decreasing response speeds and increasing costs. You can choose the appropriate model based on your specific use case and budget considerations.

For detailed model service pricing, see Billing Rules.

2. Text-to-Image Models​

LinkAI provides text-to-image models through drawing plugins, currently supporting OpenAI's DALL-E model and GPT-Image-1 model. You can enable the corresponding plugin in your agent or select it in a workflow plugin node. Learn more in the Plugins section.

3. Image Recognition Models​

LinkAI's image recognition models are also provided through plugins, currently supporting OpenAI, Claude, and Qwen image recognition models, as well as OCR image recognition. Enable the image recognition plugin in your agent and configure the desired model, or select and configure it in a workflow plugin node. Learn more in the Plugins section.

4. Voice Models​

These models are used for voice interactions on the LinkAI web platform and integrated channels. Speech recognition models are system defaults, while speech synthesis currently supports OpenAI.

5. Knowledge Base Embedding Models​

These models process uploaded knowledge base data and embed it into vector databases. They also vectorize user queries to match with knowledge base content. These models are built into the knowledge base functionality and don't require manual selection.