LLM Guide

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

Model Selection​

The model to be used during a conversation in SkyStudio can be selected from the message input area on the Chat screen.

To access the Chat page, use the Chat option in the left-hand panel of SkyStudio.

To start a new conversation, select + New Chat. On the conversation screen that opens, the Model Selector, located in the lower-right corner of the message input area, is used to determine which model will be used during 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 from the same area after the conversation has started. This allows you to move between models according to the requirements of different stages of a task.

Considerations When Selecting a Model​

Each model has different strengths and areas of use. Therefore, when selecting a model, the scope of the task and the expected level of performance should be taken into consideration.

The following criteria can be evaluated when selecting a model:

• Task complexity
• Required level of reasoning
• Response speed and performance balance
• Processing volume
• Need to work with long-form content or documents
• Coding and technical analysis requirements
• Use of multimodal content
• Tool use and agentic task requirements

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

For daily professional operations, content generation, and standard analysis tasks, models that provide a balance between speed and accuracy may be used.

For high-volume operations or tasks that require fast responses, speed-oriented models may be preferred.

GOAT Models​

GOAT Thinking​

GOAT Thinking is a general-purpose model that provides speed and efficiency for everyday enterprise tasks while being capable of applying deeper reasoning when required for more complex tasks.

The model can adapt its operating approach according to the complexity of the task. It can produce fast responses for straightforward tasks, while using an Adaptive Thinking approach to perform deeper evaluation when more comprehensive analysis is required.

USE CASES

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

TASK SUITABILITY

GOAT Thinking supports a broad range of use cases, from simple daily tasks to enterprise processes that require more advanced reasoning.

It can be particularly useful for tasks such as contract and regulatory analysis, multi-step operational decision-support processes, and the analysis of documents containing extensive industry-specific terminology.

WHEN SHOULD IT BE PREFERRED?

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

GPT Models​

GPT models can be used across a broad range of scenarios, including daily professional tasks, software development, data analysis, research, agentic workflows, and complex problem solving.

Different GPT models provide options for different requirements based on response speed, depth of reasoning, task complexity, and tool-use capabilities.

GPT-6.1 Sol​

GPT-6.1 Sol is designed for complex coding, tool use, and demanding professional tasks. It supports a range of work, from software development and research to content creation and document preparation.

USE CASES

• Complex software development
• Debugging and code review
• Completing multistep tasks using tools
• Turning technical requirements into implementation plans
• Comparing solution approaches
• Research and technical analysis
• Preparing reports and project documentation
• Everyday correspondence and extensive content creation

TASK SUITABILITY

Well suited to tasks that involve a sequence of steps, such as examining a problem, developing a solution, implementing it, and checking the results.

In addition to coding, it supports professional work such as research, detailed content preparation, and planning.

WHEN TO CHOOSE IT

For complex coding, research, or substantial projects that require multiple tools.

GPT-6 Sol​

GPT-6 Sol is designed for complex coding tasks and multistep agent workflows. It supports software development, content creation, analysis, and everyday professional work.

USE CASES

• Developing features that affect multiple files
• Examining code dependencies
• Refactoring code
• Investigating the causes of errors
• Improving code based on test results
• Developing software using external tools
• Everyday content creation and planning

TASK SUITABILITY

Well suited to software tasks that progress through a series of steps, such as reviewing a codebase, identifying the required changes, implementing them, and checking the results.

It also supports planning and content preparation beyond software development.

WHEN TO CHOOSE IT

For work that requires connected development steps or several operations to be completed together.

GPT-6 Luna​

GPT-6 Luna is designed for clearly defined tasks repeated at scale. It is suited to short everyday text tasks, classification, and information extraction.

USE CASES

• Drafting short emails and messages
• Editing text and producing brief summaries
• Categorizing support requests
• Extracting specific information from text
• Converting short content into a standard format
• Creating responses from predefined templates
• Editing product descriptions

TASK SUITABILITY

Well suited to tasks where the input, required operation, and expected result are defined in advance.

It supports applying the same operation to many pieces of content. Try a few examples before processing a larger batch.

WHEN TO CHOOSE IT

For short everyday tasks and repetitive text operations governed by clear rules.

GPT-6 Astra​

GPT-6 Astra is one of OpenAI's most capable models developed for complex professional tasks. It can be used for comprehensive workloads involving reasoning, software development, research, computer use, and document generation. OpenAI positions the model particularly for the most demanding end-to-end workflows.

USE CASES

• Complex professional tasks
• Advanced software development
• System architecture
• Research
• Technical analysis
• Computer and tool use
• Document generation
• Multi-step workflows
• Agentic tasks
• Analysis of large codebases
• Complex problem solving

TASK SUITABILITY

It is suitable for tasks that require not only analysis, but also planning, the use of necessary tools, and the end-to-end execution of multiple stages.

It can be used in professional workflows involving long context, complex instructions, and the coordinated use of multiple tools.

WHEN SHOULD IT BE PREFERRED?

GPT-6 Astra can be preferred for complex tasks that require the combination of advanced analytical capabilities, strong coding performance, research, and tool use.

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
• Analysis of large codebases
• Analysis of long documents
• Multi-step engineering processes
• Complex problem solving

TASK SUITABILITY

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

It is suitable for situations where large amounts of information must be evaluated together, technical problems must be analyzed, and tasks involving multiple stages must be completed.

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
• Daily professional tasks

TASK SUITABILITY

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

It is suitable for workflows where different tasks such as coding, analysis, content generation, and research need to be carried out within the same working process.

WHEN SHOULD IT BE PREFERRED?

It can be preferred when a balanced model is required in terms of speed and analytical capability.

It is a suitable option when a general-purpose model is needed for daily professional use.

GPT-5.6 Luna​

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

USE CASES

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

TASK SUITABILITY

It is suitable for short to moderately complex tasks where speed is the primary requirement.

It can be used in scenarios involving large numbers of similar operations or where fast responses must be provided to users.

WHEN SHOULD IT BE PREFERRED?

It can be preferred for tasks where producing results quickly is more important than performing deep analysis.

GPT-5.5​

GPT-5.5 is a general-purpose model that can be used to perform complex, multi-step real-world tasks.

USE CASES

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

TASK SUITABILITY

It can be used in processes where a task needs to be planned, divided into multiple stages, and completed using the necessary tools.

It is suitable for the end-to-end execution of tasks involving 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
• Analysis of large codebases
• Long-document analysis
• Root cause analysis
• Algorithm design

TASK SUITABILITY

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

WHEN SHOULD IT BE PREFERRED?

It can be preferred for tasks where 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

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

TASK SUITABILITY

It can be used in daily scenarios where different tasks such as coding, writing, analysis, planning, and technical support need to be performed together.

WHEN SHOULD IT BE PREFERRED?

It can be preferred for tasks that do not require the depth of advanced reasoning models but demand more comprehensive performance than standard high-speed 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 generation
• 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 need to 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 primary requirements.

GPT-5.2​

GPT-5.2 is a powerful model developed for complex professional work and tasks requiring high accuracy. Its reasoning level can be configured, allowing it to operate at different levels of depth depending on task complexity.

USE CASES

• Analysis of long and complex documents
• Technical research
• Professional reporting
• Planning
• Data analysis
• Decision-support processes
• Coding
• Multi-step problem solving

TASK SUITABILITY

It can be used for professional tasks requiring multiple pieces of information to be evaluated together with a high level of accuracy.

It is suitable for situations where long documents need to be analyzed, research findings need to be combined, and comprehensive plans need to be created.

WHEN SHOULD IT BE PREFERRED?

It can be preferred when more comprehensive analysis or professional work is required beyond standard daily tasks.

GPT-5.1​

GPT-5.1 provides strong performance particularly for coding and agentic tasks, with a configurable reasoning level. OpenAI positions the model for coding and agentic workloads.

USE CASES

• Software development
• Analysis of large codebases
• Refactoring
• Debugging
• Tool-using agents
• Agentic workflows
• Technical analysis
• Technical documentation

TASK SUITABILITY

It can be used for tasks where multiple operations need to be performed on code, tools must be used, or long-running software development processes need to be executed.

WHEN SHOULD IT BE PREFERRED?

It can be preferred when coding, refactoring, and tool-assisted agentic workflows are the primary requirements.

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 capabilities, tool use, and multi-stage 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
• Daily professional tasks
• Routine content generation
• Data analysis

TASK SUITABILITY

It is suitable for scenarios 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 for high-volume operations.

GPT-4.1​

GPT-4.1 can be used for tasks requiring long context and strong analytical capabilities.

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 that require long documents to be read, understood, and analyzed.

It is suitable for technical content generation and reporting scenarios where deep reasoning is not required and low latency is important.

WHEN SHOULD IT BE PREFERRED?

It can be preferred particularly for analysis and reporting processes involving long-form content and documents. For tasks requiring multi-step reasoning, models with stronger reasoning capabilities may be more appropriate.

OpenAI o-Series Models​

OpenAI's o-series models are reasoning models developed for tasks that require more logical reasoning and multi-stage problem solving than standard text generation. Current GPT-5.x and GPT-6 models also include built-in reasoning capabilities.

o4 Mini​

o4 Mini combines fast response times with reasoning capabilities.

USE CASES

• Customer support systems
• Dynamic question-and-answer applications
• Short- and medium-length content generation
• Rapid prototyping
• Coding examples
• Intermediate-level 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 for scenarios where both fast responses and reasoning capabilities are required.

o3​

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

USE CASES

• Scientific work
• Code analysis
• Complex algorithms
• Multi-stage analyses
• Engineering problems
• Advanced research

TASK SUITABILITY

It can be used for complex problems in fields such as mathematics, engineering, and science where detailed reasoning is required.

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 requiring fast responses and lighter reasoning.

USE CASES

• Short logical operations
• 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 primary requirements.

WHEN SHOULD IT BE PREFERRED?

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

Claude Models​

Claude models can be used for tasks involving long documents, professional content generation, software development, analysis, tool use, and complex reasoning.

Claude Fable 5.1​

Claude Fable 5.1 is a powerful model developed for advanced reasoning and long-running agentic tasks.

The model extends the capabilities of Fable 5 for long-running agentic coding, multi-step research, and document, spreadsheet, and presentation workflows. It supports a context window of 1 million tokens and a maximum output of 128,000 tokens.

USE CASES

• Long-running agentic coding
• Complex software development
• Multi-step research
• Deep technical analysis
• Analysis of large documents
• Spreadsheet workflows
• Presentation generation
• Long-running tool-use workflows

TASK SUITABILITY

It is suitable for workloads consisting of many steps where the model needs to preserve task context over an extended period.

It can be particularly useful for tasks requiring both strong reasoning and long-running agentic execution.

WHEN SHOULD IT BE PREFERRED?

It can be preferred when highly complex reasoning or long-running agent operations are required and faster models are not sufficient.

Claude Fable 5​

Claude Fable 5 is a model developed for advanced reasoning and long-running agentic workloads. Anthropic positions the model as delivering next-generation intelligence for long-running agents.

USE CASES

• Complex reasoning
• Long-running agentic tasks
• Software development
• Multi-stage research
• Technical analysis
• Tool use
• Large-scale project and document workflows

TASK SUITABILITY

It can be used for comprehensive workflows where the model must follow many steps and continue working on a task over an extended period.

WHEN SHOULD IT BE PREFERRED?

It can be preferred for workloads requiring high analytical depth and long-running execution rather than simple or short operations.

Claude Opus 5.5​

Claude Opus 5.5 is the first model in Anthropic's Claude 5.5 family. It is designed for long-running agentic coding and enterprise knowledge work.

Anthropic states that the model performs at the level of Claude Fable 5.1 on most tasks and surpasses it in areas such as agentic coding, knowledge work and computer use. It completes the same work at lower cost and faster than Opus 5. The model supports a 1 million token context window and up to 128K output tokens.

USE CASES

• Agentic coding
• Advanced software development
• Performance optimization
• Code Review
• Refactoring
• Long-running agent tasks
• Computer use
• Enterprise research
• Document and spreadsheet work

TASK SUITABILITY

It can be used for long-running professional tasks in which multiple tools, steps or agents are managed together.

It is well suited for technical processes that require high accuracy and for enterprise workflows made up of many steps.

WHEN SHOULD IT BE PREFERRED?

Claude Opus 5.5 can be preferred as a faster and more cost-efficient option for coding, agentic workflows and in-depth professional analysis that require near Fable-level quality. For short and simple tasks, Claude Sonnet models may be a better fit.

Claude Opus 5​

Claude Opus 5 is an advanced Claude model that provides strong performance for complex agentic coding and enterprise workloads.

Anthropic states that Claude Opus 5 improves upon previous Opus generations in areas such as deep reasoning, long-running agentic tasks, large refactoring operations, Code Review, and multi-agent coordination. The model supports a context window of 1 million tokens and a maximum output of 128,000 tokens.

USE CASES

• Advanced software development
• Agentic coding
• Large-scale refactoring
• Code review
• Bug detection
• Long-context analysis
• Multi-agent coordination
• Enterprise research
• Document and spreadsheet workflows

TASK SUITABILITY

It can be used for comprehensive professional tasks where multiple tools, agents, or processing steps need to be managed together.

It is suitable for technical workflows requiring long-running execution and high accuracy.

WHEN SHOULD IT BE PREFERRED?

Claude Opus 5 can be preferred when high quality is required for coding, agentic workflows, and deep professional analysis.

Claude Sonnet 5​

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

USE CASES

• Email and corporate communications
• 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 scenarios involving long conversations and documents.

WHEN SHOULD IT BE PREFERRED?

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

Claude Opus 4.8​

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

USE CASES

• Strategic analysis
• Financial analysis
• Legal analysis
• Technical analysis
• Analysis of large codebases
• Debugging
• Agentic development
• Complex research

TASK SUITABILITY

It can be used for tasks requiring planning, evaluation, deep analysis, and multi-stage problem solving.

WHEN SHOULD IT BE PREFERRED?

It can be preferred for processes that require long-running work and comprehensive analysis rather than short and simple tasks.

Claude Opus 4.7​

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

USE CASES

• Complex research
• Strategy
• Finance
• Legal analysis
• Technical analysis
• Detailed reporting
• Large-codebase analysis
• Architectural review
• Agentic development

TASK SUITABILITY

It can be used for workflows requiring planning, interpretation, analysis, and multi-stage 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 Opus 4.6​

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

USE CASES

• Analysis of large codebases
• Complex debugging
• Long-running engineering processes
• Autonomous tasks
• Tool use
• Enterprise decision support
• Complex planning

TASK SUITABILITY

It can be used for tasks that require planning, analysis, evaluation, and multi-stage technical problem solving to be performed together.

WHEN SHOULD IT BE PREFERRED?

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

Claude Sonnet 4.6​

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

USE CASES

• Code Review
• Codebase analysis
• Detection of logic errors
• Security reviews
• Testing processes
• Data analysis
• Long-document analysis
• Repetitive quality-control processes

TASK SUITABILITY

It can be used for professional processes requiring strong analytical capabilities alongside fast and repeatable execution.

WHEN SHOULD IT BE PREFERRED?

It can be preferred particularly for daily software development, code review, testing, and document analysis processes.

Claude Opus 4.5​

Claude Opus 4.5 can be used for comprehensive tasks requiring high accuracy and deep reasoning.

USE CASES

• Deep technical analysis
• Software engineering
• Complex debugging
• Long-document analysis
• Research
• Strategic analysis
• Multi-step problem solving

TASK SUITABILITY

It can be used for operations where high accuracy is expected and the model must evaluate large amounts of information together.

WHEN SHOULD IT BE PREFERRED?

It can be preferred when analytical quality and comprehensive evaluation are more important than response speed.

Claude Sonnet 4.5​

Claude Sonnet 4.5 provides strong performance for coding, tool use, and complex agent tasks while remaining a balanced option for everyday professional use.

USE CASES

• Software development
• Long-running coding tasks
• Tool-using agents
• Data analysis
• Instruction following
• Technical documentation
• Problem solving
• Enterprise workflows

TASK SUITABILITY

It can be used for processes where coding, analysis, and tool use need to be performed together within the same task.

WHEN SHOULD IT BE PREFERRED?

It can be preferred for daily professional tasks that require strong coding and analytical capabilities without the heavier processing requirements of Opus-level models.

Gemini Models​

Gemini models provide options for a variety of use cases, including advanced reasoning, multimodal content analysis, long-context management, coding, research, and high-volume processing.

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

Gemini 3.8 Flash​

Gemini 3.8 Flash is Google's most capable Flash model. It is designed for long-running software development, autonomous agents and complex enterprise workflows.

Google states that Gemini 3.8 Flash offers clear improvements over Gemini 3.7 Flash in software development, agentic tasks and multi-step reasoning in specialized domains. The model has a 1 million token context window and can process text, image, audio and video content.

USE CASES

• Long-running software development
• Agentic coding
• Tool use
• Autonomous agents
• Enterprise workflows
• Multi-step reasoning
• Multimodal analysis
• Long document analysis
• High-volume automation

TASK SUITABILITY

It can be used for tasks that require strong coding and agentic capabilities together with fast responses.

It is well suited for workflows that involve many tool calls, coding steps or different content types.

WHEN SHOULD IT BE PREFERRED?

Gemini 3.8 Flash can be preferred for tasks that do not require Pro-level heavy analysis but do call for strong coding and agentic capabilities.

Gemini 3.7 Flash​

Gemini 3.7 Flash is a Flash model built for coding and agentic tasks, offering a balance between speed and analytical capability.

Google states that Gemini 3.7 Flash improves on Gemini 3.6 Flash in software development, web development and document processing. The model has a 1 million token context window and can process text, image, audio and video content.

USE CASES

• Coding
• Debugging
• Web development
• Agentic planning
• Tool use
• Document analysis
• Workflow automation
• Multimodal analysis
• Everyday professional tasks

TASK SUITABILITY

It can be used for professional tasks where a balance of speed, cost and agentic capability matters.

It is well suited for coding and automation processes of moderate complexity.

WHEN SHOULD IT BE PREFERRED?

Gemini 3.7 Flash can be preferred for coding and automation tasks where speed and efficiency are the priority and the additional analytical depth of Gemini 3.8 Flash is not needed.

Gemini 3.6 Flash​

Gemini 3.6 Flash is a production-ready Flash model that provides a balance between speed, multimodal capabilities, and agentic workloads.

Google states that Gemini 3.6 Flash provides improved token efficiency compared with previous Flash models, along with stronger coding and agentic planning capabilities.

USE CASES

• Coding
• Agentic planning
• Tool use
• Data processing
• Multimodal analysis
• Daily professional tasks
• High-volume workflows
• Automation
• Fast technical analysis

TASK SUITABILITY

It can be used for general agentic and professional scenarios where both strong analytical capabilities and high response speed are required.

It can also be used in workflows involving a large number of tool calls or coding steps.

WHEN SHOULD IT BE PREFERRED?

It can be preferred when a fast, capable Gemini model with agentic capabilities is required without the need for Pro-level heavy analysis.

Gemini 3.5 Flash-Lite​

Gemini 3.5 Flash-Lite is a lightweight model designed for high-volume automation tasks requiring low latency and low cost.

Google positions the model as a low-latency and cost-efficient option particularly for high-volume automation and sub-agent use cases.

USE CASES

• High-volume automation
• Sub-agent tasks
• Data classification
• Data extraction
• Short summarization
• Simple content generation
• Fast question answering
• Batch data processing
• Repet

Gemini 3.1 Pro​

Gemini 3.1 Pro can be used for tasks requiring multi-source research, comprehensive analysis, and advanced reasoning across different data types. It can evaluate different types of data such as text, images, audio, and video together to perform comprehensive analysis.

USE CASES

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

TASK SUITABILITY

It is suitable not only for producing strong responses, but also for workflows that require tools to be used correctly, multi-step tasks to be executed carefully, and more consistent results to be produced.

It can be used particularly for agents, research assistants, analysis bots, and decision-support systems, as well as for multi-source research, comparative analysis, and detailed reviews.

WHEN SHOULD IT BE PREFERRED?

Gemini 3.1 Pro can be preferred for tasks that require deep analysis, tool use, multi-source research, and the combined evaluation of different content types.

Gemini 3 Flash​

Gemini 3 Flash is designed for use cases that require fast responses, low latency, and high processing volumes.

While providing the speed and cost advantages of a Flash-class model, it also brings advanced reasoning capabilities to fast-paced workflows.

USE CASES

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

TASK SUITABILITY

Gemini 3 Flash is suitable for scenarios where balancing speed and cost is critical.

It can be preferred in systems that receive a large number of requests, require fast responses, and need to keep operational costs under control.

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

WHEN SHOULD IT BE PREFERRED?

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

Gemini 2.5 Pro​

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

USE CASES

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

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

Qwen Models​

Qwen models support coding, document review, content creation, and multistep tasks. Reasoning, tool use, and image understanding capabilities vary by model.

Qwen 3.8 Flash​

Qwen 3.8 Flash is suited to content preparation, coding, and office tasks. It works with different types of content by processing images alongside text.

USE CASES

• Everyday correspondence and content preparation
• Creating presentation outlines from documents
• Turning meeting notes into action plans
• Writing explanations for reports and tables
• Identifying requirements from screenshots
• Interpreting images
• Writing Excel macros and small automation scripts

TASK SUITABILITY

Well suited to tasks that combine documents, images, and technical content.

Examples include creating an action plan from meeting notes, turning a report into a presentation, or reviewing a screenshot to identify the work required.

WHEN TO CHOOSE IT

When office work, content preparation, and coding come together in the same task.

Moonshot (Kimi) Models​

Moonshot (Kimi) models support coding, document review, and detailed analysis. Advanced reasoning and image understanding capabilities vary by model.

Kimi K3​

Kimi K3 is designed for extended coding projects, detailed analysis, and tasks requiring advanced reasoning. It also supports detailed content preparation and planning alongside projects that develop over several stages.

USE CASES

• Preparing detailed content and work plans
• Developing software in stages
• Reviewing related documents together
• Bringing research findings together
• Examining assumptions in complex problems
• Developing comprehensive solution plans
• Identifying inconsistencies between analysis results

TASK SUITABILITY

Well suited to work that revisits earlier findings in later stages and combines information from different sources into a coherent conclusion.

It supports projects where research progresses in stages, documents are compared, and solution plans are refined as new information becomes available.

WHEN TO CHOOSE IT

When work progresses through several stages and requires detailed review, comprehensive planning, or advanced reasoning.

DeepSeek Models​

DeepSeek models support text generation, coding, and technical problem solving. Reasoning, agent workflows, and image understanding capabilities vary by model.

DeepSeek V4.1 Flash​

DeepSeek V4.1 Flash is designed for text generation, coding, and agent tasks. It supports workflows that process successive requests and execute multistep operations.

USE CASES

• Drafting everyday text and responses
• Proposing solutions to technical questions
• Reviewing code and text
• Checking intermediate steps in a workflow
• Revising technical explanations
• Interpreting data returned by tools
• Extracting information from charts, tables, and screenshots

TASK SUITABILITY

Well suited to tasks that require repeated text generation, technical checks, or interpretation of tool results within a workflow.

It supports workflows that repeatedly review code, text, and tool results.

WHEN TO CHOOSE IT

For everyday text tasks or coding and agent workflows that require repeated model calls.

Agent and image capabilities are limited to the tools and content types supported in SkyStudio. The amount of content that can be processed in a single request depends on the model and SkyStudio usage limits.

Choosing a Model for Your Needs​

The model you choose in SkyStudio depends on the work you need to complete and the result you want. Models are not limited to a single use case. Models that excel at coding or analysis can also support everyday correspondence and content creation.

The lists below offer example options to help you get started. Consider response quality, writing style, speed, and usage cost together when making your choice.

Everyday Use and Content Creation​

You can work with different model families for tasks such as writing emails and messages, editing text, summarizing, translating, developing ideas, planning your day, and preparing business content.

SKYMOD

• GOAT Thinking

OPEN AI

• GPT-6.1 Sol
• GPT-6 Sol
• GPT-6 Luna
• GPT-5.6 Sol
• GPT-5.6 Terra
• GPT-5.6 Luna
• GPT-5 Mini

ANTHROPIC

• Claude Sonnet 5
• Claude Sonnet 4.6
• Claude Sonnet 4.5
• Claude Opus 5.5
• Claude Opus 5
• Claude Fable 5.1

GEMINI

• Gemini 3.8 Flash
• Gemini 3.7 Flash
• Gemini 3.6 Flash
• Gemini 3.1 Pro
• Gemini 2.5 Flash

QWEN

• Qwen 3.8 Flash

MOONSHOT (KIMI)

• Kimi K3

DEEPSEEK

• DeepSeek V4.1 Flash

Try lighter models for short correspondence, simple edits, and repetitive text tasks. Consider more capable options for detailed content, creative work, document interpretation, and comprehensive planning.

You do not need to use the same model for every everyday task. Compare the results and choose a model whose writing style and approach suit your needs.

Software Development​

Consider a model's coding and problem-solving capabilities for tasks such as writing code, debugging, refactoring, code review, and system design.

OPEN AI

• GPT-6 Astra
• GPT-6.1 Sol
• GPT-6 Sol
• GPT-5.6 Sol
• GPT-5.5
• GPT-5.1

ANTHROPIC

• Claude Opus 5.5
• Claude Opus 5
• Claude Fable 5.1
• Claude Sonnet 5

GEMINI

• Gemini 3.8 Flash
• Gemini 3.6 Flash
• Gemini 2.5 Pro

QWEN

• Qwen 3.8 Flash

MOONSHOT (KIMI)

• Kimi K3

DEEPSEEK

• DeepSeek V4.1 Flash

Producing a short code example and modifying a large codebase have different requirements. As the work becomes more complex, the model's ability to understand relationships between files, follow your instructions, and complete steps consistently becomes more important.

Review generated code and verify changes with the relevant tests.

In-Depth Analysis and Reasoning​

The following models provide starting points for solving complex problems, comparing alternatives, examining cause-and-effect relationships, and conducting multistage assessments.

OPEN AI

• GPT-6 Astra
• GPT-6.1 Sol
• GPT-6 Sol
• GPT-5.6 Sol
• GPT-5.4
• o3

ANTHROPIC

• Claude Fable 5.1
• Claude Fable 5
• Claude Opus 5.5
• Claude Opus 5

GEMINI

• Gemini 3.1 Pro
• Gemini 2.5 Pro

MOONSHOT (KIMI)

• Kimi K3

Depth and consistency matter more than speed in these tasks. Ask the model to explain its assumptions, how it compared the alternatives, and the basis for its conclusion.

Check the results against the accuracy of the information you provided and any gaps that remain.

Long Document Analysis​

When working with contracts, technical documents, academic material, extensive reports, or related documents, consider both content length and the need to establish connections between documents.

OPEN AI

• GPT-6 Astra
• GPT-6.1 Sol
• GPT-5.6 Sol
• GPT-5.2
• GPT-4.1

ANTHROPIC

• Claude Fable 5.1
• Claude Opus 5.5
• Claude Opus 5
• Claude Sonnet 5

GEMINI

• Gemini 3.8 Flash
• Gemini 3.1 Pro
• Gemini 2.5 Pro

QWEN

• Qwen 3.8 Flash

MOONSHOT (KIMI)

• Kimi K3

Summarizing, locating specific information, comparing documents, and providing detailed interpretation are different tasks. Choose a model based on what you need from the document, not just its length.

Ask the model to identify the relevant section or page so you can check important information easily.

Fast and Batch Tasks​

Speed and consistent results are key considerations for classification, short summaries, information extraction, customer support, and similar repetitive tasks.

OPEN AI

• GPT-6 Luna
• GPT-5.6 Luna
• GPT-5.4 Nano
• GPT-5 Mini
• o3 Mini

GEMINI

• Gemini 3.5 Flash-Lite
• Gemini 3 Flash
• Gemini 2.5 Flash

DEEPSEEK

• DeepSeek V4.1 Flash

Start with lighter models when the input, rules, and expected result are clearly defined. Try a more capable option for workflows that require technical interpretation or several steps.

Before processing a larger batch, compare result quality, response time, and usage cost on a few example tasks.

Research and Technical Work​

Consider a model's research and analysis capabilities when reviewing multiple sources, comparing information, combining findings, and preparing comprehensive reports.

OPEN AI

• GPT-6 Astra
• GPT-6.1 Sol
• GPT-5.6 Sol
• GPT-5.5

ANTHROPIC

• Claude Fable 5.1
• Claude Opus 5.5
• Claude Opus 5

GEMINI

• Gemini 3.1 Pro
• Gemini 2.5 Pro

MOONSHOT (KIMI)

• Kimi K3

Go beyond summaries in research tasks. Ask the model to identify differences between sources, missing information, and areas of uncertainty.

Work that requires current information needs web search or access to relevant data sources in SkyStudio. For research involving images, audio, or video, use options that support those content types.

Agent and Multistep Tasks​

Some tasks require planning, using different tools, checking intermediate results, and completing steps in sequence. Consider the model's ability to carry the work through from start to finish.

OPEN AI

• GPT-6 Astra
• GPT-6.1 Sol
• GPT-6 Sol
• GPT-5.6 Sol
• GPT-5.5
• GPT-5.1

ANTHROPIC

• Claude Fable 5.1
• Claude Fable 5
• Claude Opus 5.5
• Claude Opus 5
• Claude Sonnet 5

GEMINI

• Gemini 3.8 Flash
• Gemini 3.6 Flash
• Gemini 3 Flash

QWEN

• Qwen 3.8 Flash

MOONSHOT (KIMI)

• Kimi K3

DEEPSEEK

• DeepSeek V4.1 Flash

These tasks are not limited to software development. Research, document preparation, reporting, and office work may also involve coordinating several steps.

Consider how closely the model follows your instructions, whether its steps remain consistent, and whether it continues to account for earlier information throughout a longer task.

You can switch to another model in the Model Selector as your needs change. For example, after completing an extensive review, you can use a lighter model for summarizing and editing the text.