ToolRobustBench - Tool Interface: leaderboard
Metric: Success rate (0-1, native; interface-level perturbations to tool names, descriptions and boundaries; over ToolRobustBench single-family perturbation instances (clean, light, medium, heavy); binary per-instance tool-call success). Source: arxiv.org. Saturation forecast: Estimated already saturated. 7 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Sonnet 4.6 | 0.97 |
| 2 | DeepSeek V4 Flash | 0.96 |
| 3 | Qwen Plus | 0.93 |
| 4 | DeepSeek V4 Pro | 0.93 |
| 5 | Gemini 2.5 Pro | 0.9 |
| 6 | Gemini 2.5 Flash | 0.89 |
| 7 | GPT-5.4 Mini | 0.86 |
Interactive version: theaggregate.ai/benchmark?slug=toolrobustbench-tool-interface · How It Works · Data refreshed daily, snapshot 2026-09-26.