Skip to main content

​
Chat Layout Overview

The Chat Interface is where you interact with AI models, manage prompts, compare results across models, upload files, and organize or share conversations. Chatinterfacemain Pn The main elements include:

​
Model Selector

Choose the AI model you want to interact with. You can switch model at any time without losing context. Modelselector PN

​
Chat Input Box

Type prompts, upload files, and send instructions from the input area at the bottom of the screen. Messageinputbox Pn

​
Attachments

Upload PDF or image files (up to 50MB). Uploaded files appear inside your chat history and count as part of the conversation. Attachmentsbutton Pn

​
Chat Settings

Located at the bottom of the message input box. Chatsettings Pn Includes:
  • Model parameters like temperature (higher temperature means more creative)
  • Context window size (memory retention)
  • System instructions for the current chat (anything specific you wish the LLM to remember)
Modelsettings Pn The Chat Search feature allows you to locate any conversation within your current workspace or project. It saves time when you have a large number of chat threads and need to jump back to a specific discussion. The search bar appears at the top of the chat sidebar, labeled “Search chats”. Chatsearchfeature Pn How Chat Search Works
  • Start typing a keyword in the Search chats bar.
  • Results appear instantly and filter down the visible chat list.
  • Search matches against chat titles and keywords within each conversation thread.
Chatsearchfeatureworking Pn

​
Chat History

The chat history in the left sidebar shows all chats in the current workspace or project. You can rename, move to project, delete, or share conversation threads from here. Chathistorysidebar Pn

​
Projects

Allows you organize related chats into projects, have its own specific instruction - like a context based on client, campaign etc. Read more about Projects. Projectsfeature Pn

​
Workspace Indicator

Visible on the top right; shows your current organization and workspace and lets you switch between them. Workspaceindicator Pn

​
Stop Button

During response generation, a Stop button appears in the prompt box. Stopbutton Pn Clicking it immediately stops the model’s response. This is useful when:
  • The model starts going off-track
  • You want to refine or correct the prompt
  • You only needed a partial answer

​
Switching Models During Chat

You can change the model at any point in the same conversation thread without starting a new thread. Switching models:
  • Does not reset context
  • Continues the same conversation with all previous messages included
Switching is simple. Just go to the model selector and choose another model before sending the next prompt on the same thread.

​
Multi-Model Chat

Geekflare AI supports multi chat mode:

​
Parallel Responses (Multi-Chat)

Send the same prompt to up to three models simultaneously and view their outputs side-by-side. Select the multi chat mode, select upto 3 models from the model selector and enter your prompt. Parallelresponses Pn You get the responses side-by-side. Parallelresponsessidebyside Pn

​
Manual Switching

You can switch the model and click the ‘regenerate’ option within the same thread to compare how models handle context differently. As shown below, I have asked Gemini a question and got a response. Manualswitching Pn Now I switch the model and ask GPT 5.1 Instant the same question. Manualswitching2 Pn You can view both the responses. Viewingbothresponses Pn

​
Sharing Chats

Chats can be shared in multiple ways depending on your plan. To share, click the 3 dots next to your conversation thread name and select ‘share conversation’. Sharingconversations Pn You are presented with 3 options. Sharingconversationoptions Pn
  1. Anyone with the link: Generate a publicly viewable link accessible to anyone. This option is available in all plans.
  2. Share with Organization members: Make the chat viewable by everyone in the org (Business plan and above).
  3. Share with Specific People: Invite specific members by entering their email address (Business plan and above).
For options 2 and 3, the shared conversation will be visible to the other members under ‘shared with me’. Sharedwithme Pn

​
Permissions

  • All shared chats are view-only
  • To continue or modify the chat, the viewer must fork it and create their own copy
  • Shared chats cannot be edited directly
When you share a conversation by selecting a ‘anyone with the link’, you get a publicly shared link. You can choose stop sharing the thread when you wish. Sharingpubliclink Pn If you receive a shared chat thread or are invited to view a publicly shared link, open the thread and select the ‘fork’ option. Select the workspace into which you would like to fork this conversation to and continue chatting as usual. Forkingfromsharedconversation Pn Select ‘tools’ and turn on the web search option below the prompt box. When you type in a prompt with this feature turned on, you will receive web search results in the response. Use this feature when you need recent data and don’t want to be limited by the model’s training cutoff date. Websearchresults PN You can click ‘view all’ to view all the pages on the right side. Websearchresultsviewall PN

​
How to perform Deep Research?

Use Deep Research models offered on Geekflare AI. Currently the Deep Research models offered are: OpenAI’s o3-Deep Research, o4 mini Deep Research and Perplexity’s Sonar Deep Research. The Deep Research Settings option will be visible below the prompt box when you select a Deep Reasearch model.

​
Deep Research with OpenAI

In the case of OpenAI models, you have the option to filter domains in your result, that is a list of domains from which you want the research to be performed. Open A Ideepresearchsettings PN

​
Deep Research with Perplexity Sonar

In the case of Perplexity, you have multiple settings. Core settings: Search Mode, Disable web search, enable search classifier, reasoning effort, and whether to include images and related questions in the result. Search filters: Recency, thoroughness, user location, data filters and domain filters. Perplexitysonardeepresearchsettings PN Perplexity also offers advanced settings for deep research. Temperature: Lower value gives more stable and consistent responses, while higher values give more creative and varied responses. Top P: This refers to the range of selection of the ‘next word’. A lower number results in consistent results and higher number results in creative and unusual results. The latter is good for brainstorming. Top P and Temperature go hand-in-hand. Top K: It refers to how many next words should the model think about. A low number indicates limited and more likely options. A higher number indicates a larger pool of choices. Perplexitysonardeepresearchadvancedsettings PN Presence penalty: This tells the model not to repeat the same topic. If you set a lower value, you may find the model talking about the same thing repetitively. If you set a higher value, the model will introduce new topics. Frequency penalty: This tells the model not to repeat the same words. A lower value results in the model repeating the same words and a higher model results in the model using a wider range of vocabulary. Max Tokens: This limits the model’s maximum number of tokens used for responding. You can set it to ‘auto’ (let the model decide) or specify the number of tokens. Please note that if the Max Tokens is set at a very low value, the response may be terminated prematurely.

​
How to Create Mermaid Diagrams

Type your prompt and ask the model to output a Mermaid diagram code block. Example: 
Create a mermaid diagram showing steps in a simple email drip campaign.
Mermaid1 PN You can directly view the mermaid diagram on Geekflare AI. You can also copy/paste the Mermaid code wherever needed and paste them into tools that support diagram rendering. Emaildripcampaign PN