Gemma 4: Byte for byte, the most capable open models

Google DeepMind releases Gemma 4, an open-source model family in four sizes (E2B, E4B, 26B MoE, 31B Dense) under Apache 2.0 license.

• •31B model ranks #3 on Arena AI's open leaderboard for advanced reasoning, agentic workflows, and code generation
• •Features native vision/audio processing, 128K-256K context windows, and support for 140+ languages
• •Optimized for mobile devices, consumer GPUs, and cloud deployment via Vertex AI, Cloud Run, and GKE
• •Enables fine-tuning for specific use cases with demonstrated success in language models and scientific research

This summary was automatically generated by AI based on the original article and may not be fully accurate.