Presets & Model Tiers¶
GloriaMundo uses presets to control which AI models are assigned to each stage of the workflow pipeline. A preset is a bundle of model assignments -- one per pipeline role -- optimised for a particular balance of speed, quality, and cost.
Preset Overview¶
| Preset | User-Facing Name | Description |
|---|---|---|
AUTO | Auto | Default. Routes between Quick and Smart at runtime based on task complexity. |
QUICK | Quick | Fastest and most economical. Good for simple, well-defined tasks. |
SMART | Smart | Balanced quality and speed. Good default for most workflows. |
ULTIMATE | Ultimate | Maximum quality. Uses the most capable models at every stage. Best for complex, high-stakes workflows. |
CUSTOM | Custom | User-defined model assignments per pipeline role. |
Legacy names
Economy, Balanced, and Performance are legacy preset names that still work as aliases in the API but are no longer shown in the UI. Economy maps to Quick, Balanced maps to Smart, and Performance maps to Smart.
Auto (Default)¶
Auto mode automatically routes between Quick and Smart based on the complexity of your request. Simple, well-defined automations get the speed of Quick; complex multi-step workflows get the reasoning power of Smart.
This is the recommended preset for most users.
Quick¶
Designed for speed and low cost. Quick uses lightweight, fast models throughout the pipeline.
Best for:
- Simple, straightforward automations
- Rapid prototyping and iteration
- Workflows with clear, well-defined steps
Trade-offs:
- Less sophisticated reasoning for complex tasks
- May need more manual refinement for nuanced workflows
Smart¶
The primary AI-powered tier. Smart uses mid-to-high-tier models with reasoning capabilities.
Best for:
- Real workflows that need intelligent planning
- Multi-step automations with data transformation
- Workflows involving content generation or analysis
Trade-offs:
- Slower than Quick
- Higher cost per workflow
Ultimate¶
Maximum thoroughness for critical workflows. Ultimate uses premium models with deep reasoning for every stage of the pipeline.
Best for:
- High-stakes automations where accuracy is essential
- Complex, multi-step workflows with many dependencies
- Workflows that need the best possible content generation
Trade-offs:
- Significantly slower than other tiers
- Highest cost per workflow
- Defaults to deep research mode (20 search results), which may increase latency and cost; search depth can be changed in Settings
Custom¶
Custom mode lets you assign a specific model to each pipeline role individually. This is useful if you want fine-grained control -- for example, a fast model for classification but a premium model for content generation.
Configure custom model assignments in Settings > AI Models.
Pipeline Roles¶
Each preset assigns a model to every pipeline role. The table below describes what each role does.
| Role | What It Does |
|---|---|
strategizer | Strategic analysis and classification of the user's request |
architect | Workflow structure and step design |
solver | Step execution and problem-solving |
writer | User-facing content generation in Virtual Run |
auditor | Quality auditing of generated workflows |
coder | Code step generation and debugging |
research | Web research and information gathering |
twitter_search | X/Twitter search via Grok models |
image_model | AI image generation model selection |
tool_agent | Tool-use agent for dynamic action resolution |
Specific model names are not listed here because they change as better models become available. Each preset assigns different models optimised for the appropriate cost, speed, and quality tradeoffs for that tier. You can inspect the current assignments in Settings > AI Models.
Per-Step Model Override¶
For LLM steps, you can override the model used for a specific step without changing your overall preset. This lets you use a premium model for one critical step while keeping the rest at a lower tier.
Configure per-step model overrides in the step editor.
Changing Your Preset¶
In the Builder¶
Select a preset from the toolbar before generating a workflow.
In Settings¶
Go to Settings and find the AI Models section to set your default preset.
Cost Impact¶
Higher tiers cost more credits per workflow. Use Virtual Run to preview the cost before running, and check the cost breakdown in Settings to understand where credits are being spent.