Gemma 4 E4B

Gemma 4 E4B is Google's efficient 4.5B-effective-parameter multimodal model, ideal for laptops and consumer devices. It delivers a significant quality leap over its predecessor Gemma 3n E4B while maintaining a similar footprint. Supports text and vision inputs with 128K context. At Q4 it needs about 6 GB of VRAM, fitting comfortably on entry-level GPUs and 8 GB Macs.