TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

Perplexity API sucks – found a way to emulate it

1 点作者 warthog9 个月前
I have been trying out Perplexity&#x27;s API for weeks now. There is a big problem with it.<p>The results returned from the API are not the same as the actual Perplexity API returns. Most results seem not to be returning any kind of actual web data. At best, the returned results hallucinate.<p>I had to come up with a different way to solve for this.<p>1. I created an LLM agent using Langgraph 2. Added Google search (also DuckDuckGo search) as tools 3. For scraping purposes, you can optionally add a scraper as a tool<p>The difference to Perplexity in this case is that the results will load slower as the LLM is running the search in real time through a custom Google search engine. If anyone has found better alternatives, that would be welcome.<p>P.S.: Tried out Exa and Tavily APIs however the results get still pretty bad or outdated once the search query gets a bit more complex.<p>Curious if anyone has different methods that have been working better

暂无评论

暂无评论