Aleph Alpha and Mistral AI releases boost Europe's sovereign ambitions
German AI firm Aleph Alpha and French rival Mistral have both unveiled new models, strengthening Europe's bid to build sovereign AI alternatives to US and Chinese leaders. Aleph Alpha's Kolibri model, designed for mission-critical German-language use, comes as the company plans a transatlantic merger with Canada's Cohere.
Aleph Alpha and Mistral have released new AI models, raising hopes for Europe's sovereign AI ambitions. The surprise releases from the German and French champions signal the emergence of several credible European players in the global artificial intelligence race.
Aleph Alpha unveiled its Kolibri model. Mistral introduced what it calls a new state-of-the-art AI model. These developments are seen as critical for a continent seeking to reduce its reliance on technology giants from the United States and China.
Aleph Alpha's Kolibri model details
Aleph Alpha's Kolibri is a German-developed, sovereign AI model designed for mission-critical use with strong data compliance and efficiency features. The company, founded in 2019 and employing about 200 people across four German locations, released Kolibri on 3 October 2026. It is built for organisations in public administration and industry that require strict data control, documented compliance, and native German operation.
Kolibri is a mixture-of-experts model with 78 billion total parameters, but it engages only about 3 billion active parameters per token for greater efficiency. German-language text constitutes roughly 23% of its pre-training data. Aleph Alpha developed a German-optimised tokeniser that represents German text using fewer tokens, improving performance.
The model supports agentic workflows, retrieval-augmented generation, and native tool calling. In RAG applications, it is specifically trained to answer questions based on supplied documents and to abstain when evidence is insufficient. Aleph Alpha claims this reduces hallucinations in high-stakes settings. Kolibri also includes built-in reasoning controls to adjust the balance between reasoning depth and response speed.
To address legal and ethical concerns, Aleph Alpha screened Kolibri against a blocklist of more than 4.5 million URLs, including sources like the European Commission’s Piracy Watch List. The company accessed every third-party dataset individually to check licence terms, lawful sourcing, and adherence to opt-outs. A detailed technical report explains how Kolibri addresses copyright, data protection, and EU AI Act requirements. The model's licence terms and supporting documents are available on Hugging Face.
Leadership and transatlantic expansion
Aleph Alpha's leadership has announced a merger with Canadian AI firm Cohere to build a transatlantic sovereign AI company, pending regulatory approval. CEO Ilhan Scheer and co-founder Samuel Weinbach are steering the company through this shift. A definitive business agreement was signed last month.
The combined entity will operate globally as Cohere, with headquarters, research and development centres, and leadership roles split between Canada and Germany. The merger is subject to regulatory approval. Aleph Alpha states that with Kolibri's availability and the merger in progress, Europe can build its own large language models without compromising on sovereignty.
