WildToolBench - Task Accuracy: leaderboard
WildToolBench (2026) evaluates LLM tool use under realistic user behavior: 256 multi-turn scenarios over about 1,600 public APIs mix compositional tool orchestration, intentions hidden in earlier turns and switches between tool, chat and clarification requests. This board is accuracy in percent over all 1,024 user tasks, each model calling tools through its native function-call format; higher is better.
Metric: Task accuracy (%) over all 1,024 tasks, 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 2030. 50 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | Gemini 2.0 Flash (Thinking) | 61.04 | ||
| 2 | O1 | 58.79 | #403 | |
| 3 | O3 | 56.93 | #237 | |
| 4 | Claude Sonnet 4 | 56.54 | #410 | |
| 5 | Gemini 2.5 Pro | 56.25 | #276 | |
| 6 | GLM-4.5 | 56.05 | #542 | |
| 7 | DeepSeek R1 | 55.27 | #482 | |
| 8 | Claude Opus 4.1 | 54.98 | #278 | |
| 9 | GPT-4o | 54.88 | #751 | |
| 10 | toolace2-8B | 54.59 | ||
| 11 | GPT-4.1 | 54.3 | #505 | |
| 12 | Qwen 3 8B | Thinking | 54.3 | #1873 |
| 13 | Qwen 3 14B | Non-reasoning | 54.1 | #1269 |
| 14 | Qwen 3 4B | Reasoning | 53.91 | #2348 |
| 15 | Kimi K2 | 53.71 | #443 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=wildtoolbench-task-accuracy · How It Works · Data refreshed daily, snapshot 2026-11-21.