Well, you might hold your nose for a few days while Copilot helps you write the IntelliJ plugin for Code Llama. Now that would be a good hacker news post...

We've done work to set Continue up to be a JetBrains plugin—it's a matter of just a) figuring out the basics of putting a webview in a JetBrains plugin and b) implementing this class (https://github.com/continuedev/continue/blob/main/extension/...) that communicates with the server. So long story short, we're planning on this soon and welcome PRs. It's a medium-sized task.

UniteAI (https://github.com/freckletonj/uniteai) I think fits the bill for you.

This is my project, where the goal is to Unite your AI-stack inside your editor (so, Speech-to-text, Local LLMs, Chat GPT, Retrieval Augmented Gen, etc).

It's built atop a Language Server, so, while no one has made an IntelliJ client yet, it's simple to. I'll help you do it if you make a GH Issue!

at Refact we have Jetbrains plugin that works with a bunch of local code models https://github.com/smallcloudai/refact/

but for code llm to be useful , the local machine need to have some very powerful GPU

not true anymore. give codellama-7b-instruct a try. Just install Ollama. Pretty mind-blowing performance for the ram it uses and how fast it is. It's in the ballpark of chatgpt-3.5-turbo for code related stuff