Introduction

Web Search Plus is a search tool specifically developed for LLM that can intelligently filter out invalid information.

Web Search is the most important tool of LLM. LLM needs to obtain real-time information through web search. However, excessive information with low relevance to the search topic interferes with the inference quality of LLM, causing hallucination and reducing its inference speed. At the same time, it also generates a large amount of ineffective inference costs.

On the Internet, the average text part of a web page is 1500 words. Web Search Plus actively filters out content that is less relevant to the search topic, typically reducing content by about 70%.

Key Features:

• Content-cleaning and segmentation: Removes noise (ads, side info), segments content, and reranks results based on keyword and semantic relevance to minimize token usage.
• Rich functional configuration: Developers can control the search mode and the number of search results returned through rich configurations
• Multilingual support: Includes tokenization and ranking logic for English, Chinese, Japanese, and Korean.
• Transparent results: Unlike OpenAI's closed search tools, all retrieved content is visible to developers, allowing for verification.
• Forever pay as you go: Developers only need to pay for the tokens consumed by search results, and cannot pay a monthly fee. Not charged based on search volume, but pay as you go to save a lot of costs for lightweight searches for over 90% of the total search volume

The above all features can be completed with just one API interface. You only need to enter the search term, and the rest will be done by Web Search Plus. Let's Quickstart.