Back to AIBriefs
LaunchAI Models

Google launches Gemma 4 12B, an encoder-free multimodal model

Google launches Gemma 4 12B, an encoder-free multimodal modelblog.google

Runs locally on 16GB VRAM, supports native audio inputs. Performs near the 26B MoE model on benchmarks. Open under Apache 2.0 with Multi-Token Prediction drafters.

··Discuss
Jun 3, 4:04 PM
Google launches Gemma 4 12B, an encoder-free multimodal model — AIBriefs