BREAKING
Google Expands Gemma 4 Lineup
New 12B Unified Multimodal Model
0B
parameters
0
layers
0K
context window
Instruction-Tuned Benchmarks
AIME 202677.5
MMLU Pro77.2
LiveCodeBench72
MMMU Pro69.1
QAT vs MTP Tooling
QATQ4_0
Smaller memory footprint
E2B drops to ~1GB
Retains more quality
MTP Drafter~4 layers
Speculative decoding
~1.4x speedup
Ollama and MLX support
Compact Open Models Race Heats Up
AI NEWS BLITZ
Google DeepMind has broadened its open-weight Gemma 4 family with a new unified model.