Gemma 4 (E2B, E4B, 26B-A4B, 31B)

Parameters

E2B (~5.1B total), E4B (~8B total), 26B-A4B (25.2B total / 3.8B active), 31B dense (30.7B)

Context Window

128K (E2B/E4B), 256K (26B-A4B/31B)

License

Apache-2.0

Key Features

Open Gemma 4 family release positioned by Google as its most capable open models to date, focused on advanced reasoning and agentic workflows with high intelligence-per-parameter. Official launch materials present four sizes (E2B, E4B, 26B-A4B MoE, 31B dense) and emphasize broad deployment from edge/mobile to workstation/server settings, with configurable thinking mode, native function calling, structured JSON outputs, and native system prompt support. Model-card documentation highlights multimodal capabilities (text/image across all models, plus audio and video support on E2B/E4B), multilingual coverage (140+ pretraining languages), and benchmark gains across reasoning, coding, and long-context tasks versus prior Gemma generations. Release messaging also stresses a commercially permissive Apache 2.0 license and day-one ecosystem support across major inference and fine-tuning stacks.

Paper / Source

https://arxiv.org/abs/2607.02770

Official Links

• https://blog.google/innovation-and-ai/technology/developers-tools/gemma-4/
• https://huggingface.co/collections/google/gemma-4
• https://huggingface.co/google/gemma-4-31B-it
• https://arxiv.org/abs/2607.02770
• https://ai.google.dev/gemma/docs/core/model_card_4
• https://ai.google.dev/gemma/docs/core
• https://ai.google.dev/gemma/docs/gemma_4_license