4 ms·

Thanks! Unfortunately, LM Studio does not support network proxies, and Ollama does not support Qwen3. So it seems like Llama.cpp is the only solution.