DAREBench - Multimodal - Multi-Step with Tools: leaderboard
Metric: Accuracy (%). Source: arxiv.org. 27 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.8 | 61.8 |
| 2 | Claude Opus 4.6 | 48.1 |
| 3 | MiMo-V2.5 | 43 |
| 4 | GPT-5.5 | 42.4 |
| 5 | Claude Sonnet 4.6 | 42.3 |
| 6 | Qwen 3.6 Plus | 41.2 |
| 7 | GPT-5.4 | 40.3 |
| 8 | Qwen 3.5 Plus | 39.1 |
| 9 | Kimi K2.5 | 38.7 |
| 10 | Qwen 3.6 27B | 38.5 |
| 11 | GPT-5.4 Mini | 35.9 |
| 12 | MiMo-V2-Omni | 34.7 |
| 13 | Gemini 3.1 Pro (Preview) | 32.7 |
| 14 | Qwen 3.6 35B A3B | 32.7 |
| 15 | Claude Haiku 4.5 | 31.6 |
Interactive version: theaggregate.ai/benchmark?slug=darebench-multimodal-multi-step-with-tools · How It Works · Data refreshed daily, snapshot 2026-09-19.