Comment by shepherdjerred

Am I reading the Agents’ Last Exam v1 graph correctly where it's saying GPT Luna Medium performs better than High, but XHigh is better than Medium?