Google Cloud published a blog post explaining how to deploy the Meta Llama 3.2-1B-Instruct model on Cloud Run using GPUs. This post provides step-by-step instructions on how to leverage Cloud Run GPU for deploying open-source large language models (LLMs). The post also covers best practices to streamline the development process using local model testing with the Text Generation Inference (TGI) Docker image, making troubleshooting easy and boosting productivity. With Cloud Run GPU, developers benefit from the same on-demand availability and effortless scalability that they love with Cloud Run's CPU and memory, with the added power of NVIDIA GPUs. When your application is idle, your GPU-equipped instances automatically scale down to zero, optimizing your costs. The post also provides tips on how to improve cold starts using Cloud Storage FUSE. Cloud Storage FUSE allows developers to mount Google Cloud Storage buckets as a file system, significantly reducing cold start times.