# Pleias Unveils Ethically-Trained AI Reasoning Models Optimized for RAG with Integrated Citations

## Pleias Unveils Ethically-Trained AI Reasoning Models Optimized for RAG with Integrated Citations

VentureBeat reports that Pleias, an AI startup committed to ethical AI development, has launched a new set of small reasoning models specifically designed for Retrieval-Augmented Generation (RAG) applications. These models stand out due to their optimization for RAG workflows and their unique ability to provide built-in citations, ensuring transparency and verifiability in their outputs.

According to VentureBeat’s Carl Franzen, Pleias is positioning these models for seamless integration into a variety of applications, including search-augmented assistants, educational platforms, and customer support systems. The combination of small size, reasoning capabilities, and citation functionality makes them particularly well-suited for these tasks.

RAG, a technique increasingly utilized in natural language processing (NLP), enhances the performance of large language models (LLMs) by grounding them in external knowledge. This approach overcomes the limitations of pre-trained models by allowing them to access and incorporate real-time information from external sources. By optimizing their models for RAG, Pleias is enabling developers to create AI-powered applications that are both informed and accurate.

The built-in citation feature is a crucial aspect of Pleias’ offering. It directly addresses concerns about misinformation and “hallucinations” often associated with LLMs. By providing clear sources for the information used in its responses, the models enhance user trust and enable easy verification of the AI’s claims.

This launch from Pleias highlights the growing importance of ethically developed AI solutions. As AI becomes more deeply integrated into our daily lives, the need for transparent, accountable, and verifiable systems becomes paramount. Pleias’ focus on ethical training, combined with the practical advantages of its RAG-optimized models with built-in citations, positions the company as a significant player in the evolving AI landscape. Furthermore, the model’s compatibility with various existing platforms and the open-source nature (potentially Apache 2.0 license) could facilitate wider adoption and contribute to the advancement of responsible AI development.