Instruction Following on RecRM-Bench
72.66AccuracyOurs
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| Ours2026.05 | 72.66 | 72.4 | |
| LongCat-Flash-ChatThinking=false2026.05 | 64.84 | 65.19 | |
| GPT-4.12026.05 | 55.47 | 58.33 | |
| LongCat-Flash-ThinkingThinking=true2026.05 | 43.75 | 48.16 | |
| Qwen3-MaxThinking=false2026.05 | 36.8 | 40.33 | |
| Deepseek-V3.2Thinking=true2026.05 | 35.29 | 39.5 | |
| Deepseek-V3.2Thinking=false2026.05 | 30.47 | 29.94 | |
| Qwen3-MaxThinking=true2026.05 | 26.67 | 26.64 |