Event ·Multi-source

AutoResearchExam基准测试发布

Alex Dimakis团队发布了AutoResearchExam基准测试，涵盖7个研究领域，包括模型训练、数据整理和AI安全。测试中，Astra模型领先19小时后，Fable 5.1反超成为最佳模型。Qwen3.8 Max、Gemini 3.8 Flash和Grok 4.6在成本效益方面表现优异，位于帕累托前沿。

Current read

看看AI研究代理在24小时内如何解决开放性问题，以及不同模型在长期任务中的表现差异。

11 sources58° AI heatLast updated 2026/9/9 23:02:04

Evidence

11 sources

Primary sourceelvis

2026/9/9 23:02:04View source

Related source 1AI Will

2026/9/8 00:28:36View source

Related source 2lmarena.ai

2026/9/10 15:43:04View source

Related source 3orange.ai

2026/9/8 09:07:46View source

Related source 4Geek

2026/9/10 07:51:57View source

Related source 5Jerry Liu

2026/9/11 00:52:05View source

Related source 6Simon Willison’s Weblog

2026/9/11 03:27:16View source

Related source 7Binyuan Hui

2026/9/11 08:01:41View source

Related source 8DeepLearning.AI

2026/9/11 14:32:07View source

Related source 9shao__meng

2026/9/7 23:36:24View source

Related source 10a16z

2026/9/8 15:16:21View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.

Open the story