AI 导读
网页研究：在 1,266 个 BrowseComp 任务上，Computer 达到 66.7% 准确率，而 Pi 为 50.2%，Hermes 在各自搜索提供商上为 43.9%。 Portable 使用最少的实际时间和最少的 token。推理和私有文档保留在本地；只有搜索会联网。https://t.co/ebw1MjO0eN
正文
Web research: on 1,266 BrowseComp tasks, Computer reaches 66.7% accuracy vs 50.2% for Pi and 43.9% for Hermes on their respective search providers.
Portable uses the least wall time and the fewest tokens. Inference and private documents stay local; only search touches the web. https://t.co/ebw1MjO0eN
来源：@perplexity_ai · x.com