Common Questions About Model Selection

Does Dessix Support Custom Third-party Models?

Answer: No

Design Philosophy

To reduce the cognitive burden on users during usage, Dessix pre-configures the most suitable default models for different usage scenarios.

Scene-based Model Configuration

• Scene Matching: Dessix configures the most suitable default model for different usage scenarios
• Cost Optimization: Simple scenarios use economical models, complex scenarios use advanced models
• User Control: In supported scenarios, users can specify a particular model via mention

Scenes and Models

Dessix uses pre-configured default models based on different usage scenarios:

Scene-to-Model Mapping

Usage Scene	Default Model	Supports Specification	Selection Reason
Chat Conversations	Gemini 2.5 Pro	✅ Yes	Complex conversations require stronger reasoning
Custom Generation	Dynamic Selection	✅ Yes	Based on user specification or system default
Summary Generation	Gemini 2.5 Flash	❌ No	Summary tasks are simpler, use cost-effective models
Metadata Extraction	Gemini 2.5 Flash	❌ No	Structured extraction, prioritize cost-effectiveness
Chat Title Generation	Gemini 2.5 Flash	❌ No	Simple text generation, optimize response speed
Chat Suggestions	Gemini 2.5 Flash	❌ No	Lightweight task, optimize cost
Text Completion	Gemini 2.5 Flash	❌ No	Quick completion, optimize response speed

Advantages of Scene-based Configuration

• Cost Optimization: Simple scenarios use economical models, complex scenarios use advanced models
• Performance Balance: Pre-configured models based on scenario characteristics
• Seamless Experience: Automatic selection, no manual intervention needed

How to Specify a Model?

In scenarios that support model specification (Chat Conversations, Custom Generation), you can specify models through the following methods:

Method 1: Specify via Mention (Recommended)

Use the @ symbol to mention a model name in conversations.

Prerequisite

You need to install the corresponding model from the Dessix Community first. The entry is the Community button in the upper right corner (shortcut Ctrl + M).

Examples:

@claude-sonnet-4 Please help me analyze the performance issues in this code

@gemini-3-pro This is a complex reasoning task, please help me solve it

Method 2: Declare Model ID in Text

Directly state the model ID you want to use in the conversation:

Examples:

Use anthropic/claude-sonnet-4.5 to complete this task, ......

Please use google/gemini-2.5-pro to handle this problem

Method 3: Use Model Aliases

Each model has multiple aliases, allowing you to use shorter names:

Use gemini25pro to handle this task

Complete this with claude4sonnet

Note

The system does not support automatic model switching based on task type descriptions (such as "this is a programming task"). If you need to use a specific model, please explicitly specify the model ID, name, or alias.

Built-in Model List

Dessix currently includes the following high-quality models:

Gemini Series (Google)

Model ID	Name	Features
google/gemini-2.0-flash-001	Gemini 2.0 Flash	Fast response, image support, 1M context
google/gemini-2.5-flash	Gemini 2.5 Flash	Enhanced fast model, tool calling support
google/gemini-2.5-pro	Gemini 2.5 Pro	Professional version, powerful reasoning
google/gemini-3-pro-preview	Gemini 3 Pro Preview	Latest flagship, multimodal reasoning

Grok Series (xAI)

Model ID	Name	Features
x-ai/grok-4-fast	Grok 4 Fast	2M context, multimodal support
x-ai/grok-code-fast-1	Grok Code Fast 1	Specially optimized programming model

Claude Series (Anthropic)

Model ID	Name	Features
anthropic/claude-sonnet-4.5	Claude 4.5 Sonnet	Balanced performance and efficiency
anthropic/claude-opus-4.5	Claude Opus 4.5	Frontier reasoning model

DeepSeek Series

Model ID	Name	Features
deepseek/deepseek-chat-v3.1	DeepSeek V3.1	Conversation-optimized version

OpenAI Series

Model ID	Name	Features
openai/gpt-5	GPT-5	OpenAI flagship model, tool calling support

Other Models

Model ID	Name	Features
z-ai/glm-4.6	GLM 4.6	Z.AI model, 200K context

Design Considerations

Why Not Expose Model Selection?

Avoid Becoming a Playground Tool

• If model selection is exposed, Dessix might become a complex Playground tool
• Would need to explain to users which models are suitable for what tasks
• Model SOTA is constantly changing, which would increase users' cognitive burden

Reduce Cognitive Load

• Users don't need to understand the characteristics and applicable scenarios of various models
• Focus on the task itself rather than tool selection
• Simplify usage flow and improve efficiency

Transparency Needs for Advanced Users

For advanced users who need to know which models are being used:

Recommended Approach

• Use @ mention to explicitly specify models
• Use model IDs or aliases when asking questions
• Refer to the model list to see supported models

Usage Tips

• Use @model-name for quick specification
• Use full model IDs for precise specification
• Use short aliases (like gemini25pro) for efficiency

Best Practices

General Users

• Directly describe your needs and tasks
• Dessix will automatically select the default model based on the usage scenario
• Focus on result quality rather than model selection

Advanced Users

• Use @ mention for precise model control
• Specify models via ID or alias in Chat conversations
• Use model aliases for quick specification
• Understand different models' strengths, manually select when needed

Through this design, Dessix maintains ease of use while providing sufficient control for advanced users.