ToolBench - Trip Booking: leaderboard
ToolBench subtask measuring agent success on trip-booking API workflows.
Metric: Task Score. Source: huggingface.co. Saturation forecast: Estimated already saturated. 44 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4 | 96.7 | #438 |
| 2 | text-davinci-003 | 89.2 | #1146 |
| 3 | CodeLlama-34B-Instruct-hf | 89.17 | #1381 |
| 4 | CodeLlama-34B-hf | 88.33 | #1515 |
| 5 | CodeLlama-34B-Python-hf | 85.83 | #1552 |
| 6 | GPT-3.5 Turbo | 85.8 | #865 |
| 7 | CodeLlama-13B-Instruct-hf | 72.5 | #1532 |
| 8 | Llama 2 70B | 71.67 | #1201 |
| 9 | CodeLlama-7B-Python-hf | 68.33 | #1608 |
| 10 | LLaMA-65B | 65.8 | #1303 |
| 11 | CodeLlama-7B-hf | 63.33 | #1457 |
| 12 | CodeLlama-13B-Python-hf | 62.5 | #1623 |
| 13 | CodeLlama-7B-Instruct-hf | 61.67 | #1545 |
| 14 | Llama 2 13B | 53.33 | #1436 |
| 15 | starcoder | 51.7 | #1577 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=toolbench-trip-booking · How It Works · Data refreshed daily, snapshot 2026-10-20.