AI 导读
我们在统一的 ReAct 设置下对 15 种模型配置进行了基准测试：Web Search、Visit 和 Python。 Ling-3.0-flash-Fin 在来源核验上达到 82.45%——在开源权重模型中属于较强的结果之一。https://t.co/OgCo7zcqGx
正文
We benchmarked 15 model configurations under a unified ReAct setup: Web Search, Visit and Python.
Ling-3.0-flash-Fin achieves 82.45% on source verification — among the stronger results recorded across open-weight models. https://t.co/OgCo7zcqGx
来源：@AntLingAGI · x.com