ResearchBenchmarksPrompt Optimization on P2-hardFollow92DSGScoreMaestro82.22484.76287.389.838Dec 23, 2025Evaluation ResultsMethodMethodLinksDSGScoreMaestroJudge=Gemini 2.0 FlashJudge=Gemini 2.0 Flash2025.1292OPT2IJudge=Gemini 2.0 FlashJudge=Gemini 2.0 Flash2025.1290CRAFTJudge=gpt-5-nano-2025-...Judge=gpt-5-nano-2025-08-072025.1290PromptistJudge=Gemini 2.0 FlashJudge=Gemini 2.0 Flash2025.1287.3LM-BBOJudge=Gemini 2.0 FlashJudge=Gemini 2.0 Flash2025.1285.9RewriteJudge=Gemini 2.0 FlashJudge=Gemini 2.0 Flash2025.1285.5OriginalJudge=Gemini 2.0 FlashJudge=Gemini 2.0 Flash2025.1282.6