Toolathlon — leaderboard
Tool-use benchmark spanning many tool categories, testing whether agents can select, sequence, and combine tools to complete realistic tasks.
Metric: Score (self-reported). Source: benchmarklist.com. Status: saturation imminent. 18 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Fable 5 | 61.7 |
| 2 | Claude Mythos 5 | 61.7 |
| 3 | Claude Mythos Preview | 61.1 |
| 4 | Claude Opus 4.8 | 59.9 |
| 5 | Claude Opus 4.7 | 59.3 |
| 6 | Claude Opus 4.6 | 56.8 |
| 7 | GPT-5.5 | 55.6 |
| 8 | GPT-5.4 | 54.6 |
| 9 | Claude Sonnet 5 | 54.3 |
| 10 | DeepSeek V4 Pro | 52.8 |
| 11 | Nex N2 Pro | 51.9 |
| 12 | Claude Sonnet 4.6 | 49.4 |
| 13 | Gemini 3.1 Pro (Preview) | 48.8 |
| 14 | GLM-5.2 | 48.2 |
| 15 | Claude Sonnet 4.5 | 41 |
Interactive version: theaggregate.ai/benchmark?slug=toolathlon · How the rankings work · Data refreshed daily, snapshot 2026-07-22.