I don't know if this is how the OpenAI effort levels work, but for Qwen 3.8 27B the effort level really does get translated into the prompt.

https://huggingface.co/Qwen/Qwen3.8-27B/blob/main/chat_templ...

xhigh: "Reasoning effort is set to xhigh. Please think carefully through the task, validate key assumptions, consider plausible alternatives, and prioritize correctness, consistency, and clarity in the final answer."

low: "Reasoning effort is set to low. Keep your thinking brief and focused, moving directly to the conclusion without unnecessary elaboration."