Google opens frontier AI models to 200 European scientists

Google DeepMind and Google Cloud will provide up to 200 European scientific researchers with direct access to its frontier AI models, alongside $3 million in compute credits. The initiative, first reported by tech.eu, targets top researchers across academia and public institutions.

This marks a rare expansion of Google’s AI access, following months of tightly controlled rollouts. The company has limited its latest model, Gemini 4 Argon, to cybersecurity teams since early October, with only a modest broadening of access two weeks later. The new program could suggest a shift in approach, particularly in Europe, where discussions around AI development and regulation have been prominent.

The $3 million compute allocation may serve as a gesture toward supporting research, though its scale relative to broader industry practices remains to be seen. The move follows other recent developments, such as the release of EmbeddingGemma 2, which focused on accessibility for different use cases.

It is not yet clear how this initiative might evolve, including whether similar access could be extended elsewhere or to other types of models. The announcement comes at a time when questions about AI transparency and collaboration continue to shape industry conversations.

Sources: tech.eu