Self host Gemma 4: Deploy LLMs on Cloud Run GPUs
Gemma 4 Cloud run Gpu Ollama Vllm Llm deployment Google cloud Serverless Cloud storage fuse Cloud build Ci/cd Ai inference Machine learning Google cloud platform
A practical tutorial showing two methods to deploy Google's Gemma 4 LLM on Cloud Run with GPUs: Ollama (containerized model) for fast cold starts but requiring rebuilds for updates, and vLLM with Cloud Storage FUSE for model decoupling at the cost of slower first boot. Both achieve serverless scaling and automated CI/CD through Cloud Build. Ideal for developers building production AI applications on GCP.
Related videos
Gemma 4 production stack: Model Armor, ADK Agents, Tracing
Build full-stack apps with Google AI Studio, Cloud Run, and Cloud SQL
Building GPU-accelerated multi-agent apps with Google ADK and Gemma 4
Build an AI agent app with Go ADK, Cloud Run, and Flutter
Building distributed multi-agent systems
Your AI agent is forgetful. Here’s how to give it a brain.
Open Models Coding Essentials – Running LLMs Locally and in the Cloud Course
How to scale to 60 million requests (with ZERO ops team) using Cloud Run
I gave a Gemma 4 AI agent a sandbox and it taught itself physics | The Agent Factory Podcast
AlloyDB Lakehouse Federation: Unified access to BigQuery and Google Cloud Lakehouse
The ABCs of agent building
