Google Opens Gemma 4 Under Apache 2.0 with Multimodal and Agentic Capabilities

2026-04-16 · Source: InfoQ · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems · Depth: Advanced, quick

Summary

Google has released Gemma 4, a new family of open-weight models under an Apache 2.0 license, featuring 2B and 4B edge variants, a 26B Mixture-of-Experts (MoE) model, and a 31B dense model. This release introduces native video and image processing across all models, with audio input on the smaller variants, and context windows up to 256K tokens. The 31B dense model achieves an estimated LLMArena score of 1452, placing its performance in a bracket typically occupied by models three to five times its size. The 31B variant also scores 84.3% on GPQA Diamond and 80.0% on LiveCodeBench v6, nearly doubling the 42.4% of its predecessor, Gemma 3 IT 27B, in science reasoning and code generation. The models support function-calling, structured JSON output, and native system instructions for building autonomous agents.

Key takeaway

For AI Architects evaluating open-source models for commercial deployment, Gemma 4's Apache 2.0 license removes previous restrictions, making it a viable option for proprietary applications. Its strong benchmark performance, multimodal capabilities, and agentic features like function-calling and structured JSON output mean you can build sophisticated, tool-integrated AI systems without licensing concerns. Consider its broad distribution and specific optimizations for edge devices and large context windows when planning your next project.

Key insights

Gemma 4 offers multimodal, agentic capabilities and strong performance under a permissive Apache 2.0 license.

Principles

• Apache 2.0 enables unrestricted commercial use.
• Agentic capabilities enhance tool interaction and multi-step workflows.

Method

The Gemma 4 family includes dense and sparse (MoE) architectures, with edge variants optimized for mobile/IoT and larger models supporting extensive context windows up to 256K tokens.

In practice

• Use function-calling for external tool integration.
• Deploy edge models for on-device AI applications.
• Leverage 256K context for large document processing.

Topics

• Gemma 4
• Apache 2.0 License
• Multimodal AI
• Agentic Capabilities
• Function Calling

Code references

• ggml-org/llama.cpp

Best for: CTO, AI Architect, Computer Vision Engineer, AI Engineer, Machine Learning Engineer, AI Scientist

Related on AIssential

• Gemma 4 Local Test | New Open LLM King? — Gemma 4 offers strong multimodal capabilities and agentic workflow support under an Apache 2 license.
• Google announces Gemma 4 open AI models, switches to Apache 2.0 license — Gemma 4 offers optimized open-weight AI models for local and mobile use, enhancing performance and developer freedom with an Apache 2.0 license.
• Gemma 4 — Gemma 4 offers powerful, open-weight multimodal AI with strong local deployment and agentic capabilities under an Apache 2.0 license.
• Gemma 4: Byte for byte, the most capable open models — Gemma 4 offers unprecedented intelligence-per-parameter in open models, enabling advanced reasoning and agentic workflows across diverse hardware.
• Gemma, DeepMind's Family of Open Models — Omar Sanseviero, Google DeepMind — Gemma 4 offers highly capable, open, and on-device LLMs with an Apache 2 license and multimodal multilingual support.
• [AINews] Gemma 4: The best small Multimodal Open Models, dramatically better than Gemma 3 in every way — Gemma 4 advances open-source AI with multimodal capabilities, efficient architecture, and a permissive Apache 2.0 license.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by InfoQ.