Open-Ended Professional Tasks on BizBench 1.0 (Leaderboard)
57.1Final ScoreGemini Deep Research
Evaluation results
| Method | Links | ||||||
|---|---|---|---|---|---|---|---|
| Gemini Deep ResearchType=Prop., Tool=Yes2026.02 | 57.1 | 63.4 | 89.1 | 43.8 | 0.063 | 7,590 | |
| Shopping ResearchType=Prop., Tool=Yes2026.02 | 56.2 | 59.1 | 95.3 | 44.4 | 0.067 | 4,772 | |
| DeepSeek V3.2Type=Open, Tool=Yes2026.02 | 55.8 | 59.2 | 93.5 | 48.1 | 0.071 | 2,877 | |
| GPT-5.2Type=Prop., Tool=Yes2026.02 | 55.7 | 59.2 | 93.5 | 50.2 | 0.071 | 3,167 | |
| GPT-5.2Type=Prop., Tool=No2026.02 | 49 | 52.6 | 93.1 | 53.7 | 0.064 | 2,516 | |
| o4-mini Deep ResearchType=Prop., Tool=Yes2026.02 | 47.7 | 54.5 | 87.6 | 41.2 | 0.061 | 2,907 | |
| DeepSeek V3.2Type=Open, Tool=No2026.02 | 46.1 | 50.7 | 90.9 | 52.1 | 0.061 | 2,299 | |
| Gemini 3 ProType=Prop., Tool=No2026.02 | 44.8 | 52.7 | 84.8 | 50.7 | 0.063 | 1,495 | |
| GPT-4.1Type=Prop., Tool=No2026.02 | 42.7 | 54.6 | 80 | 51.2 | 0.06 | 1,380 | |
| Gemini 3 ProType=Prop., Tool=Yes2026.02 | 41.8 | 48.5 | 86.5 | 44.7 | 0.059 | 1,419 | |
| ChatGPT(Web Search)Type=Prop., Tool=Yes2026.02 | 38.8 | 46 | 84.6 | 48.6 | 0.051 | 2,244 | |
| Claude Opus 4.5Type=Prop., Tool=Yes2026.02 | 36.2 | 45.8 | 80.2 | 44.8 | 0.049 | 1,824 | |
| Qwen3-MaxType=Open, Tool=No2026.02 | 34.2 | 46 | 74 | 50.9 | 0.051 | 1,040 | |
| Qwen3-MaxType=Open, Tool=Yes2026.02 | 34 | 39.9 | 83.8 | 45.5 | 0.047 | 1,340 | |
| Claude Sonnet 4.5Type=Prop., Tool=Yes2026.02 | 32.9 | 45.5 | 73.3 | 48.1 | 0.046 | 1,975 | |
| GPT-4.1Type=Prop., Tool=Yes2026.02 | 32.7 | 40.2 | 81.4 | 45.7 | 0.046 | 1,434 | |
| Claude Opus 4.5Type=Prop., Tool=No2026.02 | 32.1 | 49.3 | 68 | 52.2 | 0.044 | 2,314 | |
| Claude Sonnet 4.5Type=Prop., Tool=No2026.02 | 29.1 | 41.8 | 71.6 | 52.2 | 0.041 | 1,585 |