> ## Documentation Index
> Fetch the complete documentation index at: https://docs.igent.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Models

Understand the AI models powering Maestro and how to configure them.

## What Models Does Maestro Use?

Maestro leverages multiple AI models to deliver its capabilities. The models are selected based on the task complexity and your session configuration.

### Primary Models

**Claude Sonnet 4.5** (Default)

* The primary model for most Maestro operations
* Excellent balance of reasoning capability and performance
* Used for code generation, analysis, and general tasks
* Handles complex multi-file changes effectively

**Claude Opus 4.5**

* Available for highest-quality reasoning
* Used for particularly complex problems
* Deeper analysis and more thorough consideration
* Can be selected via session settings

**Claude Haiku**

* Faster, more cost-effective model
* Used for simpler operations when appropriate
* Quick responses with lower compute cost
* Automatically selected for straightforward tasks

### Specialized Models

Maestro also uses specialized models for specific tasks:

* **Gemini 2.5 Flash**: Image generation (Generate Image tool)
* **Recraft v3**: Vector graphics and SVG generation
* **Perplexity**: Web search and research capabilities

## How Models Are Selected

### Automatic Selection

By default, Maestro automatically selects the appropriate model based on:

* Task complexity
* Context requirements
* Your session configuration
* Cost optimization preferences

**Simple tasks** (file viewing, basic editing):

* Uses Claude Sonnet or Haiku for efficiency

**Complex tasks** (multi-file refactoring, architectural decisions):

* Uses Claude Sonnet or Opus for deeper reasoning

**Specialized tasks** (image generation, web research):

* Automatically uses the appropriate specialized model

### Manual Configuration

You can influence model selection through session settings.

#### Via Settings

Access model configuration:

```
/settings → Model Selection
```

**Options**:

* **Default**: Automatic selection based on task (recommended)
* **Sonnet**: Prefer Claude Sonnet 4.5 for all tasks
* **Opus**: Use Claude Opus 4.5 for highest quality
* **Balanced**: Optimize for cost/performance mix

#### Session Settings UI

1. Click the settings icon (bottom-left)
2. Navigate to "Model Configuration"
3. Select your preferred model or strategy
4. Changes apply to current session

### Plan-Specific Model Selection

Different subscription plans may have different model access:

**Starter Plan**:

* Access to Claude Sonnet and Haiku
* Automatic selection between them

**Pro Plan**:

* Full access to Claude Opus
* All specialized models included
* Advanced configuration options

**Enterprise Plan**:

* Custom model configurations
* Priority access to latest models
* Fine-tuned selection strategies

Check your [billing page](reference/billing) for plan details.

## Model Capabilities

### Claude Sonnet 4.5

**Strengths**:

* Excellent code generation
* Strong reasoning for complex problems
* Efficient context handling
* Good balance of speed and quality

**Best for**:

* General software engineering tasks
* Multi-file projects
* API integrations
* Most use cases

### Claude Opus 4.5

**Strengths**:

* Deepest reasoning capabilities
* Superior problem decomposition
* Handles extreme complexity
* Highest quality outputs

**Best for**:

* Critical architecture decisions
* Complex algorithmic implementations
* Research and analysis
* When quality is paramount

### Claude Haiku

**Strengths**:

* Very fast responses
* Cost-effective
* Handles straightforward tasks well
* Lower latency

**Best for**:

* Simple edits and changes
* Quick iterations
* Prototyping
* Budget-conscious sessions

## Model Behavior Differences

### Code Quality

All models produce production-ready code, but with different characteristics:

* **Opus**: Most thorough, considers edge cases exhaustively
* **Sonnet**: Excellent quality, efficient implementation
* **Haiku**: Clean, functional code with less elaboration

### Reasoning Depth

How models approach complex problems:

* **Opus**: Multi-step reasoning, explores alternatives thoroughly
* **Sonnet**: Strong reasoning with good efficiency
* **Haiku**: Direct approach, less exploration

### Context Usage

How models utilize available context:

* **Opus**: Synthesizes information across entire context
* **Sonnet**: Efficient context utilization
* **Haiku**: Focused on immediate relevant context

## Cost Implications

Model selection affects session costs:

* **Haiku**: Lowest cost per operation
* **Sonnet**: Moderate cost, excellent value
* **Opus**: Highest cost, maximum capability

The automatic selection balances cost and capability based on task requirements.

**Cost optimization tips**:

* Use automatic selection for best cost/quality balance
* Reserve Opus for truly complex tasks
* Simple tasks automatically use efficient models
* Monitor usage in billing dashboard

## Complex Thinking Tools

Some specialized tools use extended thinking models:

### Complex Coding Tool

* Uses high-reasoning budget
* May use Opus-level reasoning
* Automatically applied for complex multi-file changes
* Significantly deeper analysis than standard generation

### Complex Reasoning Tool

* Research and analytical reasoning
* Produces comprehensive markdown reports
* Uses extended thinking time
* Best for architectural decisions and analysis

These tools are automatically invoked when Maestro determines the task requires deep reasoning.

**Availability**: Complex thinking tools may be disabled in certain configurations or plans.

## Changing Models Mid-Session

You can adjust model selection during a session:

```
/settings → Change model preference
```

**Effect**:

* Applies to subsequent turns
* Does not affect current operation
* Session history preserved

**When to change**:

* Switch to Opus for particularly challenging problem
* Switch to Haiku for rapid iteration on simple tasks
* Optimize costs while maintaining quality

## Model Updates

Maestro is continuously updated with the latest AI models:

* Automatic updates to new Claude versions
* Enhanced capabilities over time
* Backward compatibility maintained
* Transparent improvements

You don't need to do anything—updates happen automatically and improve your experience.

## Best Practices

### Let Maestro Choose

The automatic model selection is highly optimized:

* Analyzes task complexity
* Considers context requirements
* Balances cost and capability
* Updates based on usage patterns

**Recommendation**: Use automatic selection unless you have specific needs.

### When to Use Opus

Switch to Opus manually for:

* Critical production code
* Complex architectural decisions
* Security-sensitive implementations
* When quality cannot be compromised

### When to Use Haiku

Explicitly choose Haiku for:

* Simple file edits
* Quick prototypes
* Learning and exploration
* Cost-conscious sessions

### Monitor and Adjust

Use the billing dashboard to:

* Track model usage
* Understand cost breakdown
* Identify optimization opportunities
* Adjust strategy based on patterns

## Frequently Asked Questions

### Can I use GPT models?

Maestro is optimized for Claude models and uses Anthropic's API. We don't currently support GPT or other model families, but we use best-in-class specialized models for specific tasks like image generation.

### Does model choice affect features?

No—all Maestro features work with all models. The difference is in reasoning quality and cost, not capability.

### Can I set different models for different projects?

Model selection is per-session. You can configure each session independently based on project needs.

### Will model updates break my code?

Model updates improve quality and capabilities. They're thoroughly tested for compatibility and don't break existing functionality.

### How do I know which model is being used?

Model selection is generally transparent, but you can see session configuration in your settings. Detailed usage appears in the billing dashboard.

## Next Steps

Explore related topics:

* **[Settings](../getting-started/working-with-maestro#settings)**: Configure session preferences
* **[Billing](reference/billing)**: Understand costs and model pricing
* **[Tools](../getting-started/tools)**: Learn about specialized tools and their models
* **[Best Practices](reference/best-practices)**: Optimize your Maestro usage