Which Gemma model do you want next?

In a notable move towards community-driven development, the team behind Google's open-source Gemma language models has taken to social media to crowdsource priorities for its next release. Hugo, a developer relations engineer on the Gemma team, posted a public poll on X asking the community, "Which Gemma model do you want next?" The poll presents several high-level directions, including models with longer context, improved multilingual capabilities, better reasoning, or enhanced coding performance. This direct outreach signals a strategic shift to involve the open-source user base early in the roadmap planning process.

The Gemma family, which includes the 2B and 7B parameter models Gemma 2 and Gemma 1.1, has gained traction as a capable and efficient alternative to larger proprietary models. By opening the floor to requests, Google is likely aiming to align its development resources with the most pressing needs of its developer community, which uses Gemma for everything from on-device applications to research. The response will help determine whether the team focuses on scaling parameters, optimizing for specific tasks like translation or code generation, or improving core architectural efficiencies in the upcoming Gemma 3 or intermediate release.