DataClawBench - Easy: leaderboard

The 131 easy tasks of DataClawBench (2026), exploratory financial data analysis by LLM agents in the OpenClaw framework; numerical easy tasks need two or three reference steps. Scored as final-answer accuracy in percent from a GLM-5 judge; higher is better.

Metric: Final-answer accuracy (%) on the 131 easy tasks, OpenClaw agent (Python execution, file reading, web fetch and web search tools) in an isolated container with the read-only data environment, vendor-default sampling and reasoning, 1200 s per task, GLM-5 judge against the gold answer with sub-questions scored separately; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 8 models tracked. Updated 7 Oct 2026.

Top models

#ModelSettingScoreOverall rank
1Claude Opus 4.6High76.8#113
2Gemini 3.1 Pro (Preview)High67.6#99
3MiniMax-M2.762.6#539
4GLM-545.8#269
5DeepSeek V3.2Thinking45#419
6Qwen 3.5 Plus45#242
7Kimi K2.541.2#266
8GPT-5.4Non-reasoning23.1#135

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=dataclawbench-easy · How It Works · Data refreshed daily, snapshot 2026-11-21.