Google Announces Gemma 4 Open Model in AICore Developer Preview
Google has officially announced the release of Gemma 4, its latest state-of-the-art open AI model, available now through the AICore Developer Preview. This new model serves as the foundational technology for the upcoming Gemini Nano 4, ensuring that code written today will seamlessly function on future Android devices equipped with Gemini Nano 4 later this year. Gemma 4 is designed to deliver efficient on-device inference across the Android ecosystem, featuring native support for over 140 languages and advanced multimodal capabilities that allow applications to process text, images, and audio. The model is offered in two distinct sizes to balance performance and efficiency: E4B, optimized for complex reasoning tasks, and E2B, which prioritizes maximum speed and low latency, operating three times faster than the E4B variant. Compared to previous versions, Gemma 4 boasts up to four times faster performance and consumes up to 60% less battery power. Developers can immediately access these improvements, including enhanced chain-of-thought reasoning and conditional statement handling, to build more capable and localized AI experiences for global audiences.
Wire timeline
Google Announces Gemma 4 Open Model in AICore Developer Preview
Google has officially announced the release of Gemma 4, its latest state-of-the-art open AI model, available now through the AICore Developer Preview. This new model serves as the foundational technology for the upcoming Gemini Nano 4, ensuring that code written today will seamlessly function on future Android devices equipped with Gemini Nano 4 later this year. Gemma 4 is designed to deliver efficient on-device inference across the Android ecosystem, featuring native support for over 140 languages and advanced multimodal capabilities that allow applications to process text, images, and audio. The model is offered in two distinct sizes to balance performance and efficiency: E4B, optimized for complex reasoning tasks, and E2B, which prioritizes maximum speed and low latency, operating three times faster than the E4B variant. Compared to previous versions, Gemma 4 boasts up to four times faster performance and consumes up to 60% less battery power. Developers can immediately access these improvements, including enhanced chain-of-thought reasoning and conditional statement handling, to build more capable and localized AI experiences for global audiences.
Android Developers Blog