ToolBench - Home Search: leaderboard
ToolBench subtask measuring tool-use success for home search and retrieval goals.
Metric: Task Score. Source: huggingface.co. Saturation forecast: Estimated already saturated. 44 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4 | 97 | #438 |
| 2 | text-davinci-003 | 97 | #1146 |
| 3 | CodeLlama-34B-Python-hf | 91 | #1552 |
| 4 | CodeLlama-34B-Instruct-hf | 90 | #1381 |
| 5 | CodeLlama-34B-hf | 88 | #1515 |
| 6 | CodeLlama-13B-Python-hf | 86 | #1623 |
| 7 | CodeLlama-13B-Instruct-hf | 85 | #1532 |
| 8 | LLaMA-65B | 84 | #1303 |
| 9 | Llama 2 70B | 83 | #1201 |
| 10 | CodeLlama-7B-Python-hf | 83 | #1608 |
| 11 | starcoder | 82 | #1577 |
| 12 | GPT-3.5 Turbo | 80 | #865 |
| 13 | CodeLlama-7B-Instruct-hf | 78 | #1545 |
| 14 | CodeLlama-7B-hf | 74 | #1457 |
| 15 | Llama 2 13B | 68 | #1436 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=toolbench-home-search · How It Works · Data refreshed daily, snapshot 2026-10-20.