Writing Evaluation on HelloBench-C
52.98ScoreGemini-2.5-pro
Evaluation Results
| Method | Links | |
|---|---|---|
| Gemini-2.5-pro2026.06 | 52.98 | |
| OPERABackbone=Qwen3-8B2026.06 | 38.91 | |
| Qwen3-8B-SFTMode=SFT2026.06 | 38.7 | |
| Qwen3-8B2026.06 | 35.77 | |
| OPERABackbone=Llama3.1-8B2026.06 | 30.08 | |
| Llama3.1-8B-SFTMode=SFT2026.06 | 29.68 | |
| MiniMax-M2.52026.06 | 26.8 | |
| DeepWriter-8B2026.06 | 26.33 | |
| GPT-4o-05132026.06 | 21.52 | |
| LongWriter-Zero-32B2026.06 | 21.26 | |
| Llama3.1-8B2026.06 | 6.59 | |
| LongWriter-8B2026.06 | -48.5 |