Question Answering on SearchGymBench Simple QA 1.0 (test)
75AccuracySearchGym
Evaluation Results
| Method | Links | |
|---|---|---|
| SearchGymBackbone=Qwen3-8B2026.01 | 75 | |
| SearchGymBackbone=Qwen2.5-7B2026.01 | 71.6 | |
| DeepSeek-V3.2Category=Commercial Model2026.01 | 69.4 | |
| SearchGymBackbone=Qwen3-4B2026.01 | 52.9 | |
| Kimi-k2Category=Commercial Model2026.01 | 52.6 | |
| ASearcher-Web-QwQCategory=Open-Source Baseline2026.01 | 37.7 | |
| ASearcher-Web-7BCategory=Open-Source Baseline2026.01 | 24.5 | |
| Qwen3-4BCategory=Open-Source Baseline2026.01 | 23.5 | |
| Qwen3-8BCategory=Open-Source Baseline2026.01 | 17.6 |