Chat Interface

Assistants and Topics

Assistants

Assistants are used to apply personalized settings to a selected model, such as preset prompts and parameter presets. These settings allow the chosen model to better meet your expected workflow.

The System Default Assistant presets a common parameter (no prompt). You can use it directly or go to the Agents page to find the preset you need.

Topics

Assistants are the parent of Topics. Multiple topics (i.e., conversations) can be created under a single assistant. All Topics share the assistant's parameter settings and model settings like preset words (prompts).

Buttons within the Chat Box

New Topic creates a new topic within the current assistant.

Upload Image or Document requires model support for image uploads. Document uploads will be automatically parsed into text and provided to the model as context.

Web Search requires configuring web search-related information in the settings. Search results are returned to the large model as context. See Web Search Mode for details.

Knowledge Base enables the knowledge base feature. See Knowledge Base Tutorial for details.

MCP Server enables the MCP server feature. See MCP Usage Tutorial for details.

Generate Image is not displayed by default. For models that support image generation (e.g., Gemini), you need to manually toggle it on to generate images.

Select Model switches to the specified model for the subsequent conversation, while retaining the context.

Quick Phrases requires pre-setting common phrases in the settings, which can then be invoked and directly inserted here, supporting variables.

Clear Messages deletes all content under this topic.

Expand enlarges the chat box for entering long texts.

Clear Context truncates the context available to the model without deleting the content, meaning the model will "forget" previous conversation content.

Estimated Token Count displays the estimated Token count. The four data points are current context count, maximum context count (∞ means infinite context), character count in current input box, and estimated Token count.

Translate translates the content in the current input box into English.

Conversation Settings

Model Settings

Model settings are synchronized with the Model Settings in Assistant Settings. See Assistant Settings for details.

Message Settings

Message Separator:

Uses a separator to distinguish the message body from the operation bar.

When On

When Off

Use Serif Font:

Font style toggle. You can also change the font via custom CSS.

Show Line Numbers for Code:

Displays line numbers for code blocks when the model outputs code snippets.

When Off

When On

Collapsible Code Blocks:

When enabled, code blocks will automatically collapse if the code snippet is too long.

Code Blocks Word Wrap:

When enabled, single-line code within code snippets will automatically wrap if it exceeds the window width.

Auto-collapse Thinking Content:

When enabled, models that support "thinking" will automatically collapse the thinking process after completion.

Message Style:

Allows switching the conversation interface to bubble style or list style.

Code Style:

Allows switching the display style of code snippets.

Math Formula Engine:

• KaTeX renders faster because it is specifically optimized for performance.

• MathJax renders slower but is more comprehensive, supporting more mathematical symbols and commands.

Message Font Size:

Adjusts the font size of the conversation interface.

Input Settings

Show Estimated Token Count:

Displays the estimated Token count for the input text in the input box (not the actual context consumed, for reference only).

Paste Long Text as File:

When copying and pasting a long paragraph from elsewhere into the input box, it will automatically display as a file, reducing interference when entering subsequent content.

Render Input Messages as Markdown:

When disabled, only the model's reply messages are rendered, not the sent messages.

When Off

When On

Translate by Tapping Space 3 Times:

After entering a message in the conversation interface's input box, tapping the spacebar three times in quick succession will translate the input content into English.

Target Language:

Sets the target language for the input box translate button and for translating by tapping space 3 times.

Assistant Settings

In the assistant interface, select the desired assistant name → choose the corresponding setting from the right-click menu.

Edit Assistant

Prompt Settings

Name:

Customizable assistant name for easy identification.

Prompt:

The prompt itself. You can refer to the prompt writing style on the Agents page to edit the content.

Model Settings

Default Model:

You can fix a default model for this assistant. When adding from the Agents page or duplicating an assistant, the initial model will be this model. If this is not set, the initial model will be the global initial model (i.e., Default Assistant Model).

Auto Reset Model:

When enabled - if you switch to another model during use within this topic, creating a new topic will reset the new topic's model to the assistant's default model. When this is disabled, the new topic's model will follow the model used in the previous topic.

For example, if the assistant's default model is gpt-3.5-turbo, and I create Topic 1 under this assistant, then switch to gpt-4o during the conversation in Topic 1:

If auto-reset is enabled: when creating Topic 2, the default model selected for Topic 2 will be gpt-3.5-turbo;

If auto-reset is not enabled: when creating Topic 2, the default model selected for Topic 2 will be gpt-4o.

Temperature:

The temperature parameter controls the randomness and creativity of the text generated by the model (default value is 0.7). Specifically:

• Low temperature values (0-0.3):

  • Output is more certain and focused

  • Suitable for scenarios requiring accuracy, such as code generation, data analysis

  • Tends to select the most probable words for output

• Medium temperature values (0.4-0.7):

  • Balances creativity and coherence

  • Suitable for daily conversations, general writing

  • Recommended for chatbot conversations (around 0.5)

• High temperature values (0.8-1.0):

  • Produces more creative and diverse output

  • Suitable for creative writing, brainstorming, etc.

  • But may reduce text coherence

Top P (Nucleus Sampling):

The default value is 1. The smaller the value, the more monotonous and easier to understand the AI-generated content; the larger the value, the wider and more diverse the vocabulary range of the AI's response.

Nucleus sampling affects the output by controlling the probability threshold for vocabulary selection:

• Smaller values (0.1-0.3):

  • Only considers the highest probability words

  • Output is more conservative and controllable

  • Suitable for code comments, technical documents, etc.

• Medium values (0.4-0.6):

  • Balances vocabulary diversity and accuracy

  • Suitable for general conversations and writing tasks

• Larger values (0.7-1.0):

  • Considers a wider range of vocabulary choices

  • Produces richer and more diverse content

  • Suitable for creative writing and other scenarios requiring diverse expressions

Context Window

The number of messages to retain in the context. A larger value means a longer context and consumes more tokens:

• 5-10: Suitable for general conversations

• 10: For complex tasks requiring longer memory (e.g., tasks that generate long texts step by step according to a writing outline, needing to ensure logical coherence of the generated context)

• Note: More messages consume more tokens.

Enable Message Length Limit (MaxToken)

The maximum number of Tokens for a single response. In large language models, max token is a key parameter that directly affects the quality and length of the model's generated responses.

For example, when testing if a model is connected after filling in the key in CherryStudio, you only need to know if the model returns a message correctly, not specific content. In this case, setting MaxToken to 1 is sufficient.

Most models have a MaxToken limit of 32k Tokens, but there are also 64k, or even more. You need to check the corresponding introduction page for details.

How much to set depends on your needs. Of course, you can also refer to the following suggestions.

Streaming Output (Stream)

Streaming output is a data processing method that allows data to be transmitted and processed in a continuous stream, rather than sending all data at once. This method allows data to be processed and output immediately after it is generated, greatly improving real-time performance and efficiency.

In environments like the CherryStudio client, it basically means a typewriter effect.

When disabled (non-streaming): The model generates information and then outputs the entire paragraph at once (imagine receiving a message on WeChat);

When enabled: It outputs character by character. You can think of it as the large model sending you each character as soon as it generates it, until all are sent.

Custom Parameters

Adds additional request parameters to the request body, such as presence_penalty and other fields. Generally, most people won't need this.

The top-p, maxtokens, stream, and other parameters mentioned above are among these parameters.

Filling method: Parameter Name — Parameter Type (text, number, etc.) — Value. Reference documentation: Click to go