ResearchBenchmarksMultiple-choice Question Answering on TSFU (Time Series Feature Understanding)Follow65.6AccuracyMhybrid27.84837.64947.4557.251Jun 15, 2026Evaluation ResultsMethodMethodLinksAccuracyMhybridModel=gpt-oss-120bModel=gpt-oss-120b2026.0665.6McodeModel=gpt-oss-120bModel=gpt-oss-120b2026.0663MdirectModel=gpt-oss-120bModel=gpt-oss-120b2026.0655.6random2026.0629.3