ResearchBenchmarksGeneral Knowledge on GPQA fullFollow61.8pass@1RePO49.3252.5655.859.04Feb 11, 2026Evaluation ResultsMethodMethodLinkspass@1RePOtraining_data=SuperGPQ...training_data=SuperGPQA subset (11k)2026.0261.8GRPOtraining_data=SuperGPQ...training_data=SuperGPQA subset (11k)2026.0259.2Qwen3-8Btraining_data=SuperGPQ...training_data=SuperGPQA subset (11k)2026.0258.1LUFFYtraining_data=SuperGPQ...training_data=SuperGPQA subset (11k)2026.0249.8