> ## Documentation Index
> Fetch the complete documentation index at: https://docs.relayhub.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Web Search & Reasoning

> Enhance AI responses with web search and reasoning mode

## Overview

RelayHub gives you two powerful toggles that change how the AI processes your messages: **Web Search** and **Reasoning Mode**. Each can be enabled independently per conversation, and they can be combined for maximum depth.

<Columns cols={2}>
  <Card title="Web Search" icon="globe">
    The AI searches the internet for current information before responding. Ideal for questions about recent events, current pricing, live documentation, or anything beyond the model's training data.
  </Card>

  <Card title="Reasoning Mode" icon="brain">
    The AI shows its chain of thought -- the step-by-step reasoning process it follows before giving you a final answer. Ideal for complex analysis, math, logic, and multi-step problem solving.
  </Card>
</Columns>

## Web Search

### How It Works

When Web Search is enabled, the AI does not rely solely on its training data. Before answering your question, it searches the web for relevant, up-to-date information and incorporates what it finds into its response. Sources are cited so you can verify the information yourself.

### Enabling Web Search

<Steps>
  <Step title="Open the chat settings">
    Look for the toggle controls in the chat input area or the settings bar above the conversation.
  </Step>

  <Step title="Toggle Web Search on">
    Click the **Web Search** toggle. It turns active when enabled. The AI will now search the web for every message you send in this conversation.
  </Step>

  <Step title="Ask your question">
    Send your message as usual. The AI searches the web, finds relevant results, and weaves that information into its response. You will see source links included in the answer.
  </Step>
</Steps>

### When to Use Web Search

<Tabs>
  <Tab title="Good Use Cases">
    * Current events and recent news
    * Latest software documentation or API references
    * Current pricing, availability, or market data
    * Regulatory changes or recent legal updates
    * Competitor analysis with up-to-date information
  </Tab>

  <Tab title="Not Needed">
    * General knowledge questions (history, science fundamentals)
    * Creative writing or brainstorming
    * Code generation for well-established languages and frameworks
    * Analyzing files you have already uploaded to RelayHub
  </Tab>
</Tabs>

<Tip>
  Web Search adds a small amount of latency to responses because the AI performs searches before answering. Disable it when you do not need current information to get faster responses.
</Tip>

## Reasoning Mode

### How It Works

Reasoning Mode makes the AI's thinking process visible. Instead of jumping straight to an answer, the AI shows you the intermediate steps -- how it breaks down the problem, what assumptions it considers, and how it arrives at its conclusion. The chain of thought appears in a collapsible section above the final answer.

### Enabling Reasoning Mode

<Steps>
  <Step title="Open the chat settings">
    Find the toggle controls in the chat input area.
  </Step>

  <Step title="Toggle Reasoning on">
    Click the **Reasoning** toggle. When active, the AI will show its thought process for every response in this conversation.
  </Step>

  <Step title="Review the chain of thought">
    After sending a message, you will see the AI's reasoning steps in a collapsible block, followed by the final answer. Click to expand or collapse the reasoning section.
  </Step>
</Steps>

### When to Use Reasoning Mode

<Tabs>
  <Tab title="Good Use Cases">
    * Complex math or financial calculations
    * Multi-step logic problems
    * Debugging code where you want to understand the AI's analysis
    * Strategic decisions where you want to see pros and cons weighed
    * Any situation where you need to audit the AI's thought process
  </Tab>

  <Tab title="Not Needed">
    * Simple factual questions with straightforward answers
    * Quick text formatting or rewriting tasks
    * Casual conversation or brainstorming where speed matters more than rigor
  </Tab>
</Tabs>

<Note>
  Reasoning Mode may increase response length and token usage because the AI generates both the chain of thought and the final answer. The reasoning tokens are included in your usage metrics.
</Note>

## Combining Both Features

You can enable Web Search and Reasoning Mode at the same time. When combined, the AI searches the web for current information and then reasons through that information step by step before giving you an answer. This is particularly powerful for:

* Researching a topic and building an analysis from current data
* Fact-checking claims using live sources with transparent reasoning
* Making decisions that depend on both current information and careful logic

<Warning>
  Using both features together produces the longest responses and highest token usage. Use the combination when depth matters; disable one or both when speed is the priority.
</Warning>

## Settings Persistence

Web Search and Reasoning settings are saved per conversation. If you enable Web Search in one conversation, it stays enabled every time you return to that conversation -- but it does not affect your other conversations. Each conversation maintains its own independent settings.