DataClawEval - Chinese Tasks: leaderboard

Metric: Rule-based task score (%; per task a weighted sum, usually 0.7 and 0.3, of the artifact score, checked row by row against live databases by task-specific deterministic scripts, and the process score for exploration, execution efficiency and self-verification, mean over the 50 Chinese-language tasks; one run per task, every model in the same Tencent CodeBuddy agent scaffold). Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 16 models tracked.

Top models

#ModelScore
1CodeBuddy + Claude Sonnet 574.1
2CodeBuddy + Claude Opus 4.873.6
3CodeBuddy + GPT 5.572.9
4CodeBuddy + Gemini 3.5 Flash72.3
5CodeBuddy + Gemini 3.1 Pro72.1
6CodeBuddy + DeepSeek V4 Flash69.7
7CodeBuddy + MiniMax M369.4
8CodeBuddy + Kimi K2.668.3
9CodeBuddy + DeepSeek V4 Pro66.9
10CodeBuddy + GLM 5.166.6

Interactive version: theaggregate.ai/benchmark?slug=dataclaweval-chinese-tasks · How It Works · Data refreshed daily, snapshot 2026-09-29.