Educational scholarly writing on EduResearchBench
3.63Overall ScoreGPT-5
Evaluation Results
| Method | Links | |||||||
|---|---|---|---|---|---|---|---|---|
| GPT-5Model Category=Closed-source Models2026.01 | 3.63 | 4.03 | 3.12 | 3.65 | 3.96 | 3.58 | 3.63 | |
| Gemini 3 FlashModel Category=Closed-source Models2026.01 | 3.34 | 4.07 | 2.92 | 3.45 | 3.82 | 3.36 | 3.21 | |
| EduWriteModel Category=Capability-enhanced LLMs, Parameter Count=30B2026.01 | 3.2 | 3.33 | 2.87 | 3.07 | 3.72 | 3.32 | 2.74 | |
| InnoSpark-72BModel Category=Capability-enhanced LLMs, Parameter Count=72B2026.01 | 2.97 | 3.31 | 2.75 | 3.06 | 3.17 | 2.98 | 2.56 | |
| Qwen-2.5-72B-InstructModel Category=Open-source LLMs, Parameter Count=72B2026.01 | 2.9 | 3.41 | 2.72 | 3.01 | 3.28 | 2.62 | 2.42 | |
| Llama-3-70B-InstructModel Category=Open-source LLMs, Parameter Count=70B2026.01 | 2.46 | 2.5 | 2.51 | 2.67 | 2.81 | 2.34 | 1.5 | |
| WritingBenchModel Category=Capability-enhanced LLMs, Backbone=Qwen2.5-7B2026.01 | 2.44 | 2.4 | 2.19 | 2.33 | 3 | 2.62 | 1.71 | |
| LongWriterModel Category=Capability-enhanced LLMs, Backbone=GLM4-9B2026.01 | 2.34 | 2.62 | 2.27 | 2.55 | 2.43 | 2.26 | 1.77 |