Skip to main content
TR

LLM Guide

SkyStudio Model Guide​

One of SkyStudio's core values is providing users with a wide range of AI models rather than limiting them to a single option. Each model has its own strengths, and we encourage you to explore which model best fits your needs.

Model Selection​

In SkyStudio, the model used during a conversation is selected from the message input area on the Chat screen.

You can access the Chat page by selecting Chat from the SkyStudio left-hand navigation panel.

To start a new conversation, select + New Chat. On the chat screen, the Model Selector located in the lower-right corner of the message input area is used to choose the model for the conversation.

The Model Selector displays the name of the currently active model. To change the model, click the Model Selector and choose the model you want to use from the list.

After selecting a model, enter your message to start the conversation.

You can also switch to a different model after the conversation has started. This allows you to use different models at different stages of a task according to your requirements.

Considerations When Selecting a Model​

Each model has different strengths and use cases. For this reason, the scope of the task and the expected model performance should be considered when making a selection.

The following criteria can be evaluated when selecting a model:

  • Task complexity
  • Required level of reasoning
  • Response speed
  • Processing volume
  • Need to work with long content or documents
  • Coding and technical analysis requirements
  • Multimodal content requirements
  • Balance between speed and performance

For tasks that require deep analysis and multi-step problem solving, models with stronger reasoning capabilities may be preferred.

For everyday professional tasks, content generation, and standard analysis, models that provide a balanced combination of speed and accuracy can be used.

For high-volume operations or use cases that require fast responses, speed-oriented models may be more suitable.

GOAT Models​

GOAT Thinking​

GOAT Thinking is a general-purpose model designed to provide speed and efficiency for everyday enterprise tasks while using deeper reasoning when more complex tasks require it.

The model can adapt its behavior according to task complexity. It can respond quickly to simple tasks while applying a deeper evaluation through an Adaptive Thinking approach for tasks that require more comprehensive analysis.

Use Cases​

GOAT Thinking can be particularly suitable for the following scenarios:

  • Daily operational tasks
  • Marketing and sales content
  • Human resources content
  • Corporate announcements
  • Email creation
  • FAQ content
  • Short reports
  • Repetitive text generation
  • Document analysis
  • Document summarization and creation
  • Operational decision-support processes

Task Suitability​

GOAT Thinking can be used across a broad range of scenarios, from simple everyday tasks to enterprise processes that require more complex reasoning.

It can be particularly useful for tasks such as contract and regulation analysis, multi-step operational decision-support processes, and analyzing documents that contain extensive industry-specific terminology.

When Should It Be Preferred?​

GOAT Thinking can be preferred when both fast daily tasks and more comprehensive analysis need to be handled by the same model.

GPT Models​

GPT models can be used across a wide range of scenarios, from everyday professional tasks to software development, data analysis, research, and complex problem solving.

Different GPT models provide options for different requirements in terms of speed, reasoning depth, and task complexity.

GPT-5.6 Sol​

GPT-5.6 Sol can be used for complex tasks that require deep analysis and high accuracy.

Use Cases​

  • Software development
  • System architecture
  • Data analysis
  • Academic work
  • Technical research
  • Large codebase analysis
  • Long document analysis
  • Multi-step engineering processes
  • Complex problem solving

Task Suitability​

It can be used for complex problems where deep reasoning and high accuracy are the primary requirements.

It is suitable for scenarios that require evaluating large amounts of information together, analyzing technical problems, and completing tasks with multiple stages.

When Should It Be Preferred?​

It can be preferred when depth of analysis and accuracy are more important than response speed.

It is particularly suitable for comprehensive technical analysis, system design, research, and complex engineering tasks.

GPT-5.6 Terra​

GPT-5.6 Terra provides balanced performance across speed, accuracy, and general-purpose use.

Use Cases​

  • Software development
  • Data analysis
  • Technical documentation
  • Research
  • Reporting
  • Project tracking
  • Corporate content generation
  • Everyday professional tasks

Task Suitability​

It can be used for both routine tasks and operations with moderate complexity.

It is suitable when coding, analysis, content generation, and research need to be performed within the same workflow.

When Should It Be Preferred?​

It can be preferred when a balanced model is needed between speed and analytical capability.

It is a suitable option when looking for a general-purpose model for everyday professional use.

GPT-5.6 Luna​

GPT-5.6 Luna can be used for tasks that require fast responses and high processing volume.

Use Cases​

  • Email writing
  • Text editing
  • Summarization
  • Translation
  • Customer support
  • Daily office tasks
  • Chatbot use cases
  • Repetitive content generation
  • Short question-answer tasks

Task Suitability​

It is suitable for short- and medium-complexity tasks where speed is the priority.

It can be used in scenarios where many similar operations need to be completed or users expect quick responses.

When Should It Be Preferred?​

It can be preferred when fast output is more important than deep analysis.

GPT-5.5​

GPT-5.5 is a general-purpose model that can be used for complex and multi-step real-world tasks.

Use Cases​

  • Software development
  • Research
  • Data analysis
  • Document creation
  • Table creation
  • Agentic coding
  • Tool use
  • Multi-step professional tasks

Task Suitability​

It can be used in processes where a task must be planned, divided into stages, and completed using the required tools.

It is suitable for carrying out end-to-end tasks that involve multiple operations.

When Should It Be Preferred?​

It can be preferred for tasks that go beyond a single question-and-answer interaction and require multiple steps or tools to be used together.

GPT-5.4​

GPT-5.4 can be used for technical and analytical tasks that require deep reasoning.

Use Cases​

  • Scientific analysis
  • Advanced engineering
  • System architecture
  • Financial modeling
  • Decision-support processes
  • Large codebase analysis
  • Long document analysis
  • Root cause analysis
  • Algorithm design

Task Suitability​

It can be used for tasks that require comprehensive evaluation, such as architectural decisions, algorithm design, system analysis, and multi-layered error analysis.

When Should It Be Preferred?​

It can be preferred when deep analysis and comprehensive problem solving are more important than speed.

GPT-5.4 Mini​

GPT-5.4 Mini can be used for professional tasks that require a balance between speed and analytical capability.

Use Cases​

  • Everyday professional use
  • Intermediate-level coding
  • Data analysis
  • Long conversations
  • Backend development
  • Frontend development
  • Script creation
  • Debugging
  • Refactoring

Task Suitability​

It can be used in everyday scenarios where coding, writing, analysis, planning, and technical support tasks are performed together.

When Should It Be Preferred?​

It can be preferred for tasks that do not require the depth of advanced analysis models but need more comprehensive work than standard fast models.

GPT-5.4 Nano​

GPT-5.4 Nano is designed for fast and lightweight operations.

Use Cases​

  • Short questions
  • Simple code generation
  • Summarization
  • Translation
  • Note creation
  • Text editing
  • Small scripts
  • Short analyses

Task Suitability​

It is suitable for tasks that do not require deep analysis, can be completed quickly, and may be repeated at high volume.

When Should It Be Preferred?​

It can be preferred for simple and repeatable operations where speed and efficiency are the priority.

GPT-5.3​

GPT-5.3 provides a balance between general-purpose use and advanced reasoning.

Use Cases​

  • Everyday professional use
  • Technical analysis
  • Software development
  • Technical research
  • Multimodal content analysis
  • Code analysis
  • Refactoring
  • Debugging

Task Suitability​

It can be used for professional scenarios that include standard tasks as well as tasks that occasionally require more comprehensive analysis.

When Should It Be Preferred?​

It can be preferred when everyday workflows require both fast operations and a certain level of technical analysis.

GPT-5​

GPT-5 is a general-purpose model that can be used for advanced reasoning, agentic tasks, and comprehensive technical operations.

Use Cases​

  • Advanced coding
  • Agentic tasks
  • Tool use
  • Debugging large codebases
  • Feature development
  • Text and image analysis
  • Enterprise workflows
  • Research and analysis

Task Suitability​

It can be used for complex tasks that require strong analytical capability, tool use, and multi-step execution.

When Should It Be Preferred?​

It can be preferred for comprehensive tasks that combine coding, research, and tool use.

GPT-5 Mini​

GPT-5 Mini is the GPT-5 family model designed for fast, high-volume operations.

Use Cases​

  • High-volume question answering
  • Classification
  • Summarization
  • RAG-based search
  • Everyday professional tasks
  • Routine content generation
  • Data analysis

Task Suitability​

It is suitable for use cases where well-defined and repeatable tasks need to be completed quickly.

When Should It Be Preferred?​

It can be preferred when speed and efficiency are important in high-volume workloads.

GPT-4.1​

GPT-4.1 can be used for tasks that require long context and strong analytical capability.

Use Cases​

  • Technical support
  • Scientific reporting
  • Analytical reporting
  • Long-form content generation
  • Education
  • Research
  • Document analysis
  • Document summarization

Task Suitability​

It can be used for tasks where long documents need to be read, understood, and analyzed.

It is suitable for scenarios that require multi-step reasoning and technical content generation.

When Should It Be Preferred?​

It can be preferred especially for analysis and reporting workflows that involve long content and documents.

OpenAI Reasoning Models​

OpenAI Reasoning models can be used for tasks that require more logical reasoning and multi-step problem solving than standard text generation.

o4-mini​

o4-mini combines fast response times with reasoning capability.

Use Cases​

  • Customer support systems
  • Dynamic question-answer applications
  • Short- and medium-length content generation
  • Rapid prototyping
  • Coding examples
  • Intermediate reasoning tasks

Task Suitability​

It can be used for tasks where low latency is important but a certain level of reasoning is also required.

When Should It Be Preferred?​

It can be preferred in scenarios where both fast responses and reasoning capability are required.

o3​

o3 can be used for tasks that require strong logical reasoning and multi-step problem solving.

Use Cases​

  • Scientific studies
  • Code analysis
  • Complex algorithms
  • Multi-step analyses
  • Engineering problems
  • Advanced research

Task Suitability​

It can be used for complex problems in fields such as mathematics, engineering, and science that require detailed reasoning.

When Should It Be Preferred?​

It can be preferred when logical accuracy and depth of problem solving are more important than speed.

o3-mini​

o3-mini can be used for operations that require fast responses and lighter reasoning.

Use Cases​

  • Short logical tasks
  • Everyday chatbots
  • Data classification
  • Simple text generation
  • Lightweight reasoning tasks

Task Suitability​

It is suitable for real-time interaction scenarios where speed and efficiency are the priority.

When Should It Be Preferred?​

It can be preferred for fast tasks that are not highly complex but still require basic reasoning.

Claude Models​

Claude models can be used for long-document workflows, professional content generation, software development, analysis, and tasks that require complex reasoning.

Claude 5 Sonnet​

Claude 5 Sonnet is a general-purpose model that can be used for professional content generation, software development, document analysis, and other tasks.

Use Cases​

  • Email and corporate communication
  • Report preparation
  • Technical documentation
  • Meeting summaries
  • Presentation content
  • Code generation
  • Debugging
  • Refactoring
  • Code review
  • Long document analysis

Task Suitability​

It can be used for enterprise tasks where both speed and analytical capability are important.

It is suitable for use cases that involve long conversations and documents.

When Should It Be Preferred?​

It can be preferred when content generation, document analysis, and software development tasks need to be handled in a balanced way.

Claude Opus 4.8​

Claude Opus 4.8 is a powerful model that can be used for complex research, analysis, and multi-step tasks.

Use Cases​

  • Strategic analysis
  • Financial analysis
  • Legal analysis
  • Technical analysis
  • Large codebase review
  • Debugging
  • Agentic development
  • Complex research

Task Suitability​

It can be used for tasks that require planning, review, deep analysis, and multi-step problem solving.

When Should It Be Preferred?​

It can be preferred for long-running workflows and comprehensive analysis rather than simple, short tasks.

Claude 4.7 Opus​

Claude 4.7 Opus can be used for tasks that require high accuracy and comprehensive analysis.

Use Cases​

  • Complex research
  • Strategy
  • Finance
  • Legal work
  • Technical analysis
  • Detailed reporting
  • Large codebase analysis
  • Architecture review
  • Agentic development

Task Suitability​

It can be used for processes that require planning, interpretation, analysis, and multi-step problem solving rather than single-step operations.

When Should It Be Preferred?​

It can be preferred when long documents, large datasets, or complex technical problems need to be evaluated together.

Claude 4.6 Opus​

Claude 4.6 Opus can be used for advanced software engineering and complex analysis tasks.

Use Cases​

  • Large codebase analysis
  • Complex debugging
  • Long engineering workflows
  • Autonomous tasks
  • Tool use
  • Enterprise decision support
  • Complex planning

Task Suitability​

It can be used for tasks where planning, analysis, review, and multi-step technical problem solving are required together.

When Should It Be Preferred?​

It can be preferred for comprehensive software engineering and long-running technical analysis workflows.

Claude 4.6 Sonnet​

Claude 4.6 Sonnet provides a balance between everyday professional use and strong code and document analysis.

Use Cases​

  • Code review
  • Codebase analysis
  • Logic error detection
  • Security reviews
  • Testing processes
  • Data analysis
  • Long document analysis
  • Repetitive quality checks

Task Suitability​

It can be used for professional processes that require strong analytical capability together with fast and repeatable execution.

When Should It Be Preferred?​

It can be preferred especially for everyday software development, code review, testing, and document analysis workflows.

Gemini Models​

Gemini models provide options for different use cases such as advanced reasoning, multimodal content analysis, long-context management, coding, research, and high-volume processing.

When selecting a model from the Gemini family, consider the required depth of analysis, response speed, types of data to be processed, and processing volume.

Gemini 3.1 Pro​

Gemini 3.1 Pro can be used for tasks that require multi-source research, comprehensive analysis, and advanced reasoning across different data types.

Use Cases​

Gemini 3.1 Pro can evaluate different types of data together, including text, images, audio, and video, to perform comprehensive analysis.

It is suitable for use cases such as:

  • Text and document analysis
  • Image and screenshot analysis
  • Table analysis
  • Audio recording evaluation
  • Video content analysis
  • Web information research
  • Multi-source research
  • Comparative analysis

Because it can connect to current web content through Google Search grounding, it can also be used in scenarios that require real-time research and source-grounded responses.

It is particularly suitable for workflows where information must be researched on the web, verified across different sources, and combined into a consolidated summary.

Task Suitability​

Gemini 3.1 Pro can be used not only for producing strong responses but also for workflows where tools need to be used correctly, multi-step tasks must be executed carefully, and more consistent results are required.

It is therefore suitable for use cases such as:

  • Agents
  • Research assistants
  • Analysis bots
  • Decision-support systems
  • Multi-source research
  • Comparative analyses
  • Detailed reviews

For tasks that require broad world knowledge and accuracy-focused work, it can be used where multiple sources need to be evaluated together rather than simply answering a single question quickly.

When Should It Be Preferred?​

Gemini 3.1 Pro can be preferred when deep analysis, tool use, multi-source research, and evaluation of different content types are required together.

It is particularly suitable for research and decision-support processes where comprehensive analysis and reliable output are more important than speed.

Gemini 3 Flash​

Gemini 3 Flash is designed for scenarios that require fast responses, low latency, and high processing volume.

It is positioned to provide Flash-level speed and cost advantages while offering reasoning capability close to Pro-level models.

Use Cases​

Gemini 3 Flash can be used especially for fast user interactions and real-time operations.

For example:

  • Live chat systems
  • Customer support bots
  • Real-time guidance agents
  • High-traffic applications
  • Fast classification
  • Data processing
  • Data cleaning
  • Function calling
  • Real-time analysis
  • Rapid prototyping
  • Batch data processing

It can also be used in operational processes that require organizing unstructured data, managing many function calls, and quickly interpreting image- or text-based content.

Task Suitability​

Gemini 3 Flash is suitable for scenarios where the balance between speed and cost is critical.

It can be preferred in systems that receive large numbers of requests, require fast responses, and need to manage operational cost efficiently.

It can also be used as a fast and efficient option in agentic workflows.

Because it can process different data types such as text, images, audio, code, and video quickly, it is suitable for real-time multimodal applications such as:

  • Media analysis
  • Live assistants
  • Automated classification
  • Content review
  • Fast generation workflows

When Should It Be Preferred?​

Gemini 3 Flash can be preferred when speed, low latency, and high processing volume are more important than deep analysis.

It is particularly suitable for live user interactions, high-traffic applications, and fast agentic workflows.

Gemini 2.5 Pro​

Gemini 2.5 Pro is one of the Gemini family's powerful thinking models designed for tasks that require advanced reasoning and coding.

Use Cases​

Gemini 2.5 Pro can be used especially for:

  • Complex software development
  • Multi-step logical problems
  • Technical analysis
  • Decision-support processes
  • Code review
  • Refactoring
  • Test scenario generation
  • Mathematical problem solving
  • Technical problem solving
  • Long document analysis

With a long-context window of up to 1 million tokens, the model is suitable for working with large document collections and long conversations.

This capability allows it to be used for tasks where many documents or large amounts of information need to be evaluated within the same workflow.

Task Suitability​

It can be used for tasks that require high accuracy and advanced reasoning.

It is particularly suitable for scenarios such as:

  • Code review
  • Refactoring
  • Test scenario generation
  • Mathematics
  • Technical problem solving
  • Analysis of long legal documents
  • Technical document analysis
  • Comprehensive report analysis
  • Evaluation of large data descriptions

When Should It Be Preferred?​

Gemini 2.5 Pro can be preferred for projects where the highest quality and advanced reasoning are the main priorities.

It is particularly suitable for complex coding, technical problem solving, and tasks involving long documents.

For lighter and higher-volume operations, Flash models may be preferred for speed and cost efficiency.

Gemini 2.5 Flash​

Gemini 2.5 Flash is one of the Gemini models that provides a balance between performance, speed, and cost.

With a 1 million token context window, the model can work with long content while also supporting everyday tasks that require fast responses.

Use Cases​

Gemini 2.5 Flash can be used especially for:

  • Chat systems
  • Customer support assistants
  • Summarization
  • Quality control
  • Extracting insights from data
  • Document summarization
  • Log analysis
  • Text analysis
  • Dashboard interpretation
  • Routine coding tasks

It is suitable for systems with frequently repeated queries that require near-real-time responses.

Task Suitability​

Gemini 2.5 Flash is positioned as a balanced option between performance and speed.

It can be used for chatbots serving large numbers of users, intelligent dashboard interpretation, log and text analysis, document summarization, and routine coding tasks.

It is suitable for scenarios where low latency and cost are important but deeper reasoning may also occasionally be required.

The model can provide deeper reasoning when needed through thinking mode.

When Should It Be Preferred?​

Gemini 2.5 Flash can be preferred when a balanced model is required across speed, cost, and analytical capability.

It is particularly suitable for high-volume enterprise use cases, chatbots, routine analysis processes, and everyday professional tasks.

Model Selection by Use Case​

When selecting a model in SkyStudio, you do not need to use a single model for every task. Instead, the model can be selected according to the requirements of the operation being performed.

Everyday Use and Content Generation​

For tasks such as email writing, text editing, summarization, corporate content generation, and everyday question answering, fast and general-purpose models can be preferred.

The goal in these tasks is to obtain fast and consistent results without unnecessary processing overhead.

Software Development​

For tasks such as code generation, debugging, refactoring, code review, and system design, models with strong coding and reasoning capabilities can be preferred.

As task complexity increases, using stronger reasoning models may be more appropriate.

Deep Analysis and Reasoning​

For complex engineering problems, algorithms, architectural decisions, scientific work, and multi-step analyses, models with strong reasoning capabilities can be preferred.

In these tasks, depth of analysis and accuracy may be more important than response speed.

Long Document Analysis​

When working with contracts, technical documents, academic papers, and comprehensive reports, models capable of handling long context can be used.

If many documents need to be evaluated together within the same task, the model's context capacity should be considered.

Fast and High-Volume Operations​

For classification, short summarization, data extraction, customer support, and similar repeatable tasks, fast and lightweight models can be preferred.

In these scenarios, response time and processing cost may become more important factors in model selection.

Research and Technical Work​

For tasks that require evaluating multiple sources, analyzing technical data, and preparing comprehensive reports, models with strong analytical and long-context capabilities can be used.

If different content types need to be evaluated within the same task, models with multimodal capabilities can be preferred.

Agentic and Multi-Step Tasks​

For workflows where a task must be planned, different tools need to be used, and multiple operations need to be executed sequentially, models with strong agentic and tool-use capabilities can be preferred.

Model selection is not fixed. If the requirements of a task change during a conversation, you can switch to another model using the Model Selector in the lower-right corner of the chat input area.

This allows a workflow that begins with a fast model to continue with a stronger reasoning model when more comprehensive analysis is required. Similarly, after completing the detailed analysis, you can switch to a faster model for lighter tasks such as summarization or text editing.