Continue.dev would be also my first idea. I tried out last week running a local model with ollama and lmstudio, but it didn't work well for me. Specially code completion, the load for my PC was quite high.