开源与闭源模型差距再度拉大
中文摘要
开源与闭源模型差距再度拉大，闭源模型在智能体能力上取得突破，而开源模型尚未跨越该门槛。
English Summary
The gap between open and closed AI models has widened, with closed models leading in agentic capabilities that open-source models have yet to achieve.
原文节选
从定性上看，开源模型与闭源模型之间的差距现在是许久以来最大的。Fable/Astra 级别的模型具备前 Fable 模型所没有的智能体能力，无论好坏。目前还没有任何开源模型跨过这条线。一旦跨过，将是一次跃升。