What's new in the Gemma open model family
Google launches Gemma 4 open model family with four sizes from 2B to 31B parameters
“Our Gemma model, here evaluated on L M Arena, are scoring as well as model 20X the size.”
Google DeepMind announced Gemma 4, its most capable open-weight model family to date, spanning 2B to 31B parameters for deployment from IoT edge devices to consumer GPUs. The family features a 26B mixture-of-experts variant and expands context windows from 32K to 128K-256K tokens. With 100M+ total downloads across the Gemma family, this signals Google's deepening commitment to competing in the open-weights space against Meta's Llama lineup.