Google's Gemma model family: which size is right for which task?

Gemma is Google's family of open models, and Gemma 4 is built for efficient, local, and on-device use. The lineup breaks down like this: Gemma 4 E2B is the smaller edge model designed for phone workflows. Gemma 4 E4B is the most practical starting point for most local tasks. Gemma 4 12B is a middle ground with more capability, suited for laptops. Gemma 4 26B/31B is in stronger local workstation territory. Beyond the main size picker, there are also specialized Gemma models. Embedding Gemma, for example, is built specifically for search — it helps you turn text into embeddings so you can search by meaning. If you want to search your own docs, customer notes, support tickets, sales calls, or a knowledge base, embeddings matter a lot.