The EduQwen 32B models achieve up to 96.52% accuracy on the CDPK Benchmark, surpassing the previous leaderboard leader Gemini-3 Pro at 90.55%¶

Claim · All claims
Evidence · 1 study · design r2 · q2
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person · recheck by 2029: tool and model accuracy changes quickly

Subclaims¶

q2 i? EduQwen 32B-RL1, EduQwen 32B-SFT, and EduQwen 32B-SFT-RL2 each surpass all open-source and proprietary models on the CDPK Benchmark leaderboard, including Gemini-3 Pro at 90.55%. → Singh 2026

Evidence¶

Singh 2026¶

Singh, N. P., Wang, X., Garikipati, A., Ciobanu, M., Mao, Q., & Das, R. (2026). Application-Driven Pedagogical Knowledge Optimization of Open-Source LLMs via Reinforcement Learning and Supervised Fine-Tuning. Preprint. https://arxiv.org/abs/2604.06385 doi:10.48550/arxiv.2604.06385

q2 · i? · design · r2

Benchmark evaluation of the three optimized models on the CDPK Benchmark, a publicly available collection of 920 teacher multiple-choice exam questions, scored by accuracy against a ground truth answer key. The final model reached "96.52% accuracy" versus Gemini-3 Pro's 90.55%. No effect size is printed.

"EduQwen 32B-RL1 reaches 94.13% accuracy, EduQwen 32B-SFT reaches 96.20% accuracy, and with the optional second RL round, EduQwen 32B-SFT-RL2 reaches 96.52% accuracy, with all three models surpassing all open-source and proprietary systems currently listed on The Pedagogy Benchmark Leaderboard, including the previous best-performing Gemini-3 Pro at 90.55%."

Discussion¶

Related Claims¶

• A single difficulty-weighted SFT run on a vision-language base model reaches 61.64% overall accuracy on TutorBench, surpassing Gemini-3 Pro (58.52%) — related
• The domain-specialized 32-billion-parameter model surpasses significantly larger general-purpose models such as Qwen3-235B-A22B-Thinking (86.63%) and Qwen2.5-72B-Instruct (78.60%) on the CDPK Benchmark — related
• LLM accuracy on the CDPK pedagogy benchmark ranges from 28% to 89%, with closed-weight reasoning models dominating the top of the leaderboard — related
• In the fine-tuning experiments, increasing base model size up to 32B parameters improved performance, and larger LLMs consistently outperformed smaller MLMs — related