Question Answering on SearchGymBench Complex QA 1.0 (test)
47.6AccuracyDeepSeek-V3.2
Evaluation Results
| Method | Links | |
|---|---|---|
| DeepSeek-V3.2Category=Commercial Model2026.01 | 47.6 | |
| SearchGymBackbone=Qwen2.5-7B2026.01 | 41.1 | |
| SearchGymBackbone=Qwen3-8B2026.01 | 40.6 | |
| Kimi-k2Category=Commercial Model2026.01 | 24.4 | |
| SearchGymBackbone=Qwen3-4B2026.01 | 20 | |
| ASearcher-Web-QwQCategory=Open-Source Baseline2026.01 | 17.8 | |
| ASearcher-Web-7BCategory=Open-Source Baseline2026.01 | 7.2 | |
| Qwen3-8BCategory=Open-Source Baseline2026.01 | 3.3 | |
| Qwen3-4BCategory=Open-Source Baseline2026.01 | 3.3 |