[wcallahan]

Great advice. Have you observed any other differences? I’ve been wondering if there are any specialized variants yet of GPT-OSS models yet that outperform on specific tasks (similar to the countless Llama 3 variants we’ve seen).