WildToolBench: 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 session accuracy in percent, the share of dialogues in which all four user tasks are completed correctly, each model calling tools through its native function-call format; higher is better.
Metric: Session accuracy (%): share of the 256 dialogues in which all four tasks are completed correctly, 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 2031. 50 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | Gemini 2.0 Flash (Thinking) | 14.45 | ||
| 2 | Gemini 2.5 Pro | 14.06 | #276 | |
| 3 | Claude Sonnet 4 | 12.5 | #410 | |
| 4 | O1 | 12.11 | #403 | |
| 5 | GLM-4.5 | 12.11 | #542 | |
| 6 | GPT-4o | 11.72 | #751 | |
| 7 | Claude 3.7 Sonnet | 11.33 | #470 | |
| 8 | Kimi K2 | 10.55 | #443 | |
| 9 | Grok 4 | 10.16 | #319 | |
| 10 | O3 | 10.16 | #237 | |
| 11 | Qwen 3 30B A3B | Non-reasoning | 9.77 | #1156 |
| 12 | DeepSeek R1 | 9.38 | #482 | |
| 13 | DeepSeek V3 | 9.38 | #729 | |
| 14 | Qwen 3 14B | Reasoning | 9.38 | #1269 |
| 15 | Claude Opus 4.1 | 9.38 | #278 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Example tasks
这个周末我想去外地游玩,请帮我查下周末两天Chicago的天气。
Interactive version: theaggregate.ai/benchmark?slug=wildtoolbench · How It Works · Data refreshed daily, snapshot 2026-11-21.