LangChain has now integrated the Gemini Pro API into its platform. This integration allows developers to use Gemini’s multimodal functionalities within its environment. 

Gemini, a generative AI model developed by Google, was released in the first week of December. The model stands out for its ability to process both text and image data in prompts. 

The integration of Gemini Pro API into LangChain has enabled the adaptation to its natively multimodal capabilities. LangChain has developed approaches to leverage this feature, especially in the context of Retrieval Augmented Generation (RAG) applications. 

Traditionally focusing on text, RAG applications are now expanding to include visual content, thanks to multimodal LLMs like GPT-4V. LangChain has explored methods like multimodal embeddings and multi-vector retrievers to effectively retrieve and synthesise information from both text and visual inputs, such as slide decks.

To further improve the developer experience, LangChain has launched its inaugural standalone integration package named ‘langchain-google-genai.’ This package offers direct access to the Gemini API, enhancing the ease with which developers can apply LangChain’s multimodal capabilities. 

Moreover, LangChain has introduced an integration guide to help developers fully utilise the Gemini Pro API’s potential. This collaboration and the introduction of these new resources opens up new opportunities in AI application development for enterprise customers.