Add-on Deals shopping assistance on SHOPPING COMPANION benchmark
66AccuracyGPT-5
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GPT-5Category=Closed2026.03 | 66 | 54 | |
| Qwen3-4B-LoRA + RL (Dual&Tool-wise Reward)Category=Ours2026.03 | 55 | 43 | |
| Qwen3-4B-LoRA + RL (Dual-reward)Category=Ours2026.03 | 50 | 38 | |
| Qwen3-4B-LoRACategory=Ours2026.03 | 42 | 31 | |
| GPT-4oCategory=Closed2026.03 | 41 | 26 | |
| GPT-4.1Category=Closed2026.03 | 39 | 24 | |
| Qwen3-MaxCategory=Closed2026.03 | 35 | 24 | |
| Qwen3-Next-80B-A3BCategory=Open2026.03 | 29 | 18 | |
| Qwen3-30B-A3BCategory=Open2026.03 | 21 | 13 | |
| Qwen3-4BCategory=Open2026.03 | 11 | 6 |