I haven't used GPT-OSS 120B, or other GPT-OSS models, and I mostly go on personal recommendations rather than benchmarks directly.