FastGPT API results are different (and much worse) than kagi.com/fastgpt

estheruary

Here's the results from the web app:

Here's what I get from the API:

{
  "data": {
    "output": "According to recent reports from May 2023, Taylor Swift has been dating **Matty Healy**, the frontman of the 
band The 1975.[1]",
    "references": [
      {
        "snippet": "Matty Healy (2023) Image Credit: Mauricio Santana/Getty Images On May 3, 2023, sources told The Sun that
 Taylor Swift is reportedly dating The 1975 frontman, Matty Healy. A source told the UK...",
        "title": "Who\u2019s Taylor Swift Dating Now 2023? Boyfriends Travis Kelce, Matty ...",
        "url": "https://stylecaster.com/lists/taylor-swift-boyfriend/"
      }
    ],
    "tokens": 0
  },
  "meta": {
    "api_balance": 4.258853,
    "id": "e9b227b5-286b-40e8-8143-46625c31a896",
    "ms": 0,
    "node": "us-central1"
  }
}

And it's not a fluke I can reproduce it with lot of basic questions from current events.

The FastGPT API should be better at sourcing relevant and more results from the search engine. It's clearly possible because the webapp does it. In an Ideal world it would give me the same or similar results to searching Kagi directly.