OpenHarmony Bench - Specification Tasks: leaderboard
Metric: Mean Task Completion (%). Source: arxiv.org. 8 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Kimi K3 | 34.67 |
| 2 | Qwen 3.8 Max (Preview) | 34 |
| 3 | GLM-5.2 | 32.67 |
| 4 | Qwen 3.7 Max | 30.67 |
| 5 | GLM-5.1 | 28.67 |
| 6 | MiniMax-M3 | 28 |
| 7 | Kimi K2.7 Code | 27.33 |
| 8 | DeepSeek V4 Pro | 24 |
Interactive version: theaggregate.ai/benchmark?slug=openharmony-bench-specification-tasks · How It Works · Data refreshed daily, snapshot 2026-09-19.