What Google has teased about Gemini 4

Google has confirmed that it is currently training Gemini 4, its next-generation frontier AI model. CEO Sundar Pichai stated that the company is 'being very ambitious with it' and is applying significant computational resources toward the model's development. The announcement was made during a recent update, though specific details about capabilities or release timeline were not disclosed.

Gemini 4 is expected to succeed Gemini 3, building on Google's existing line of large language models. The company has not provided benchmarks or target parameters, but the emphasis on 'significant compute' suggests a large-scale training run. Pichai's comments indicate a strategic push to maintain competitiveness in the frontier AI race.

The training of Gemini 4 comes amid ongoing discussions about AI safety and regulation, as frontier models increasingly attract scrutiny from policymakers. Google has previously committed to responsible AI development, but the full implications of Gemini 4 will depend on its eventual capabilities and deployment.