How to improve performance if possible?

Hi so im using tensorflow for nodejs and im making a bot to detect scam messages on discord and so far it works. The issue is that it uses A LOT of resources just for one message sent. Is there any ways to improve its performance by any chance? I figure there wont be much ways to improve it but wanted to ask just in case im missing out on something.

@Cookie
To enhance the performance of your TensorFlow bot for scam message detection on Discord using Node.js, consider implementing batch processing to handle multiple messages simultaneously. Opt for a smaller and more efficient model to reduce resource usage. Ensure you’re using the latest version of TensorFlow.js and leverage GPU support for hardware acceleration if available. Implement caching to avoid redundant model reloading, process messages asynchronously, and control the message processing rate with throttling. Conduct code profiling to identify and optimize bottlenecks, and explore quantization to reduce model precision for faster inference. Lastly, verify that your server has ample CPU and memory resources to handle the workload effectively.