Skip to content

Latest commit

 

History

History
128 lines (89 loc) · 3.93 KB

fastgpt.md

File metadata and controls

128 lines (89 loc) · 3.93 KB
Raw

FastGPT

FastGPT is a Kagi service using powerful LLMs to answer user queries running a full search engine underneath. Think ChatGPT, but on steroids and faster! You can try the web app here.

Quick start

  1. Get the API key (requires a Kagi account)
  2. Top off your API credits.
  3. Call the API. See example usage.

API key

To obtain an API key you will first need to create a Kagi account.

Then, navigate to Settings -> Advanced -> API portal, or click here to go directly.

Click "Generate API Token".

API Credits

The FastGPT API uses pre-paid API credits. You can manage API credits in the API billing page of your Kagi settings.

Using API with insufficient credits will produce "Insufficient credit to perform this request." response message.

Pricing

Pricing for FastGPT is a flat rate per-query:

  • 1.5¢ per query ($15 USD per 1000 queries) when web_search is enabled.
  • 0.15¢ per query ($1.5 USD per 1000 queries) when web_search is disabled.

Cached responses, as with all other Kagi APIs, are free.

Support

You can reach out through support@kagi.com, our Discord server or KagiFeedback.org.

Endpoints

Answer Query

POST /fastgpt

Answer a query.

Returns a FastGPT Answer.

Examples

POST request with web search

curl -v \
  -H "Authorization: Bot $TOKEN" \
  -H "Content-Type: application/json" \
  --data '{"query": "Python 3.11"}' \
  https://kagi.com/api/v0/fastgpt

Alternatively in python:

import requests

base_url = 'https://kagi.com/api/v0/fastgpt'
data = {
    "query": "python 3.11",
}
headers = {'Authorization': f'Bot {TOKEN}'}

response = requests.post(base_url, headers=headers, json=data)
print(response.json())
Response 
{
  "meta": {
    "id": "120145af-f057-466d-9e6d-7829ac902adc",
    "node": "us-east",
    "ms": 7943
  },
  "data": {
    "output": "Python 3.11 was released in 2021 and introduced several new features compared to previous versions:
    - Faster performance: The interpreter was optimized, resulting in up to 1.25x speedup on benchmarks.[1]
    - New syntax features: Exception groups were added using PEP 654.[1]    
    - Traceback annotation: The new version includes a feature to annotate tracebacks to make them more intelligible.[2]
    - Type hint improvements: Type hints were made more powerful.[3]",
    "tokens": 757,
    "references": [
      {"title": "What's New In Python 3.11 — Python 3.11.3 documentation", "snippet": "...", "url": "https://docs.python.org/3/whatsnew/3.11.html"}, 
      {"title": "Introducing the New Features in Python 3.11 -", "snippet": "...", "url": "https://earthly.dev/blog/python-3.11-new-features/"}
    ]
  }
}

Parameters

Field Type Required Description
query string Yes A query to be answered.
web_search bool No Whether to use web search results in answering the query. (default is true)
cache bool No Whether to allow cached requests & responses. (default is true)

Objects

FastGPT Answer

Field Type Description
output string Answer output
references array of Reference object The search results that are referred in the answer.
tokens int Amount of tokens processed

Reference object

Field Type Description
title string Title of the referenced search result.
snippet string Snippet of the referenced search result.
url string URL of the referenced search result.