Try LM Studio or llama-server: https://simonwillison.net/2025/Aug/19/gpt-oss-with-llama-cpp...