Google enables Gemma 4 31B to autonomously debug and execute code

Google

Apr 10, 2026 · Updated Jun 6, 2026

Google's Gemma 4 31B can now use an ADK Agent and a code execution sandbox to autonomously navigate complex, multi-step tasks. This update brings frontier-level agentic capabilities like self-debugging and tool use to an open-weight model that can run on-device or at the edge.

Google released Gemma 4 31B, an open-weight model (a model whose trained parameters are publicly released for local use) capable of functioning as an autonomous agent. By integrating with the ADK Agent and a code execution sandbox, the model can now perform zero-shot code generation, tool usage, and learned multimodal output.

This shift moves the Gemma family beyond text generation into Agentic AI. While autonomous debugging was previously limited to massive proprietary models, bringing these skills to a 31B parameter model allows developers to build reliable, self-correcting systems that run locally or on-edge without high API costs.

You can use Gemma 4 31B to build agents that handle ambiguous tasks, such as processing server logs, by allowing the model to write, test, and fix its own code. The model is available for local inference using tools like vLLM, providing a foundation for private, agentic workflows.

View the full update on developers.googleblog.com

Google AI Developers

@googleaidevsApr 10

💎 Gemma 4 31B can leverage an ADK Agent and code execution sandbox to autonomously navigate complex, ambiguous tasks. Follow along as this demo showcases: + Zero-shot code generation + Tool usage + Multi-step debugging and recovery + “Learned” multimodal output https://t.co/DJMqDux1qu

784

View on X

Every HeadsUpAI update is written based on its original source and reviewed before it's published. Read our editorial standards →

See all AI news & updates from Google →

Keep reading

Google Launches On-Device Agent Skills for Offline Gemma 4 Workflows

Google released the Google AI Edge Gallery app and LiteRT-LM framework to enable fully offline agentic workflows on mobile and IoT devices. By running Gemma 4 locally, developers can build multi-step agents that plan, use tools, and process multimodal data without cloud latency or privacy risks.

GoogleApr 27

Google Launches Gemma 4 to Bring Frontier Reasoning to Local Devices

Google released Gemma 4, a new family of open models built on the same architecture as Gemini 3 and licensed under Apache 2.0. These models deliver high-performance reasoning and native multimodal capabilities directly on consumer hardware, enabling private, offline agentic workflows. This shift allows developers to build sophisticated AI applications that run entirely on-device without sacrificing intelligence.

Ollama Adds Google DeepMind's Gemma 4 12B for Local Agentic AI

OllamaJun 7

Ollama Adds Google DeepMind's Gemma 4 12B for Local Agentic AI

Ollama has made Google DeepMind's Gemma 4 12B model available for local execution, including support for chat and agentic applications. This expands access to a powerful, open-weight multimodal model optimized for on-device reasoning and coding, enabling private and offline AI workflows on consumer hardware.

Vercel brings Google Gemma 4 to AI Gateway for high-performance agentic workflows

VercelApr 2

Vercel brings Google Gemma 4 to AI Gateway for high-performance agentic workflows

Vercel now supports Google's Gemma 4 models on its AI Gateway, offering native function calling and structured JSON output for building autonomous agents. These 26B and 31B models feature a 256K context window and are built on the same architecture as Gemini 3. This integration allows developers to deploy high-performance open models with enterprise-grade reliability and no price markup.