WildToolBench - Single-Tool Tasks: leaderboard
Metric: Accuracy (%) on tasks solvable with one tool call, on WildToolBench (256 multi-turn scenarios, four user tasks each, 1,024 tasks, built from real user-log patterns with expert-annotated tool calls), each model through its native function-call format with default decoding; higher is better. Source: arxiv.org. Saturation forecast: Around 2028. 57 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | O3 | 61.72 |
| 2 | GPT-4o | 60.16 |
| 3 | Claude Sonnet 4 | 60.16 |
| 4 | Qwen 3 8B (Non-reasoning) | 60.16 |
| 5 | Qwen 3 4B (Reasoning) | 60.16 |
| 6 | Doubao-1.5-Thinking-Pro | 60.16 |
| 7 | Grok 4 | 59.38 |
| 8 | Qwen 2.5 72B Instruct | 58.98 |
| 9 | DeepSeek V3 | 58.98 |
| 10 | Doubao-1.5-Pro | 58.59 |
| 11 | Claude 3.7 Sonnet | 57.81 |
| 12 | Qwen 3 32B (Non-reasoning) | 57.81 |
| 13 | GLM-4.5 | 57.81 |
| 14 | GPT-4.1 | 57.42 |
| 15 | Doubao-Seed-1.6 (Thinking) | 57.42 |
Interactive version: theaggregate.ai/benchmark?slug=wildtoolbench-single-tool-tasks · How It Works · Data refreshed daily, snapshot 2026-10-07.