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: years away from saturation. 47 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 5 (Max) | 80.6 |
| 2 | Claude Opus 4.8 (Max) | 79.9 |
| 3 | GLM-5.3 Flash | 78.4 |
| 4 | Claude Fable 5.1 (Max) | 77.8 |
| 5 | Claude Opus 5 | 77.6 |
| 6 | Kimi K3 | 76.5 |
| 7 | Muse Spark 1.1 | 75.6 |
| 8 | GPT-5.6 Sol | 74.9 |
| 9 | GPT-5.6 Terra | 74.9 |
| 10 | Claude Sonnet 5 (Max) | 74.7 |
| 11 | GPT-5.6 Luna | 74.6 |
| 12 | DeepSeek V4 Pro (0813) | 74.1 |
| 13 | GPT-5.5 (xHigh) | 73.5 |
| 14 | Qwen3.8-Flash-Next | 73.5 |
| 15 | GLM-5.3 | 73 |
Interactive version: theaggregate.ai/benchmark?slug=toolathlon · How It Works · Data refreshed daily, snapshot 2026-09-05.