Alongside the release of EU-compliant language models, German AI company Aleph Alpha has unveiled PhariaAI, an enterprise-grade operating system for generative AI. This new platform combines sovereign design with Large Language Model explainability and compliance.
The platform allows organizations to integrate AI into their core processes while maintaining a high level of transparency and control. Built to be compliant with stringent regulatory requirements, PhariaAI includes tools for explainability and auditability, giving experts the ability to oversee and adjust AI systems, ensuring that they operate in line with organizational standards.
Jonas Andrulis, founder and CEO of Aleph Alpha, stated, "PhariaAI is a game-changer for enterprises and governments seeking to leverage AI without compromising control or compliance. Our technology accelerates AI innovation while ensuring transparency, auditability, and full compliance with organizational standards."
One of the core missions of PhariaAI is to streamline the AI lifecycle, from development to full-scale production. Enterprises and public institutions can use the platform to scale AI projects quickly, thanks to its suite of developer tools, including components for fine-tuning, evaluation, and debugging. PhariaAI also offers ready-made templates for common tasks such as Q&A, Summarization, Translation, and Transcription—all designed to speed up implementation and optimize operations.
A key feature of PhariaAI is its hybrid execution capability, which supports the flexible distribution of AI workloads across both on-premise and cloud environments. This allows organizations to scale their operations seamlessly while ensuring data privacy and security are maintained in regulated environments. For businesses dealing with sensitive information, this flexibility is invaluable, allowing them to push capacity to the cloud during peak times while processing confidential data on-premise.
At the heart of PhariaAI are Aleph Alpha’s newly launched Pharia-1-LLM-7B-control and Pharia-1-LLM-7B-control-aligned models, which represent a significant leap forward from Aleph Alpha’s previous Luminous model family. These models are optimized for critical applications, offering adaptability, safety, and compliance in diverse environments. The aligned version incorporates additional safety guardrails, ensuring it can be used for sensitive conversational AI tasks.
Importantly, both models are trained to fully comply with the GDPR and the forthcoming EU AI Act, making them well-suited for deployment in enterprises and public institutions across Europe. The models support seven languages, with optimization for English, German, French, and Spanish, positioning them as a multilingual AI solution for diverse markets.
Aleph Alpha’s launch of PhariaAI is an important offering in the evolution of enterprise AI for Europeans, combining sovereignty, compliance, and scalability into one powerful platform.