Gemma 4 Runs on Your Phone

2025-09-15 · Source: unwind ai · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Robotics & Autonomous Systems, Software Development & Engineering · Depth: Intermediate, medium

Summary

Google has released Gemma 4, a new family of open-source, multimodal AI models under the Apache 2.0 license, ranging from a 2B edge model for devices like Raspberry Pi to a 31B dense model that ranks #3 on Arena AI's text leaderboard. These models, built on Gemini 3 research, are designed for agentic workflows with native function calling, structured output, and up to 256K context windows. Concurrently, Cursor 3 has launched, transforming its IDE into a unified workspace for managing fleets of coding agents across multiple repositories, featuring Composer 2, an integrated browser, and a plugin marketplace. Anthropic also introduced a "NO_FLICKER" mode for Claude Code, enhancing terminal-based coding sessions by virtualizing the viewport to eliminate screen flickering.

Key takeaway

For AI Architects evaluating model deployment strategies, Gemma 4's Apache 2.0 license and edge-native capabilities for devices like Raspberry Pi present a compelling option for cost-effective, offline agentic applications. Consider integrating these smaller, efficient models into your infrastructure to expand AI functionality beyond cloud-dependent solutions, especially for scenarios requiring low latency or privacy-sensitive local processing.

Key insights

Agentic AI models are becoming smaller, more efficient, and integrated into developer workflows.

Principles

• Edge-native models enable offline, low-latency AI applications.
• Agentic design requires built-in reasoning and structured output.
• Parallel agent management enhances developer productivity.

Method

Karpathy's workflow for self-improving wikis involves LLMs compiling raw research into interconnected Markdown files, with queries generating new content that loops back into the wiki.

In practice

• Deploy Gemma 4's 2B model on a Raspberry Pi for offline AI.
• Use Cursor 3 to manage multiple coding agents simultaneously.
• Enable CLAUDE_CODE_NO_FLICKER=1 for flicker-free Claude Code sessions.

Topics

• Gemma 4
• AI Agents
• Open-Source Models
• Coding Agents
• Edge AI

Code references

• lemonade-sdk/lemonade
• Shubhamsaboo/awesome-llm-apps

Best for: CTO, VP of Engineering/Data, AI Architect, AI Engineer, Machine Learning Engineer, Director of AI/ML

Related on AIssential

• DeepMind’s New AI: A Gift To Humanity — Gemma 4 offers high performance and broad utility in a compact, open-source package.
• 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.
• Bring state-of-the-art agentic skills to the edge with Gemma 4 — Gemma 4 enables powerful, multi-modal, agentic AI experiences directly on diverse edge devices.
• Google just dropped Gemma 4... (WOAH) — Gemma 4 offers high intelligence per parameter, enabling powerful AI on consumer and edge devices.
• Gemma 4 Tool Calling Explained: Build AI Agents with Function Calling (Step-by-Step Guide) — Gemma 4's native tool calling enables local AI agents to reliably interact with external APIs for real-world data.

Open in AIssential →

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