Skip to main content
Omnifact allows you to switch between different models to power your conversations. While the default model is capable of handling most tasks, you may find that certain models excel in specific areas, such as creative writing, coding, or deep logical reasoning.

Model Guide

Choose the right AI model for your task, from flagship reasoning models to fast, efficient options.
The available models are configured by your Team Administrator in Model Management. If you don’t see a specific model, it may not be enabled for your organization. Your administrator can also restrict availability to EU-hosted models only to ensure data residency compliance.

​
How to Switch Models

You can select the AI model before you send a message or at any point during a conversation directly in the chat input area.
1

Locate the Model Selector

In the bottom-right corner of the chat input field (next to the send button), look for the Model Selector pill button. It displays the provider logo and the name of the currently active model (e.g., “GPT-5.6 Sol” or “Claude Sonnet 5”). For reasoning models, it also shows the current reasoning effort level.
The model selector pill in the chat input showing Claude Opus 5.5 with the Medium reasoning effort, next to the dictation and send buttons
2

Open the Menu

Click the model selector to reveal the list of available models. Each model shows its provider logo, name, and hosting region flag. Hover over a model to see a short description beside the menu. For reasoning models, the Reasoning Effort row appears at the bottom of the menu.
The open model selector menu listing available AI models with provider logos and hosting region flags, with Claude Opus 5.5 selected and the Reasoning Effort row set to Medium at the bottom
3

Choose a Model

Select the model you wish to use. The selector in the input bar will update to reflect your choice, and all subsequent messages in that conversation will be processed by the selected model.

​
Configuring Reasoning Effort

Reasoning models (such as GPT-5.6 Sol, GPT-5.4, or Gemini 3.5 Flash) spend extra compute time “thinking” through problems step by step before generating an answer. Instead of jumping straight to a response, they evaluate intermediate logic, explore hypotheses, and verify constraints. When you select a reasoning-capable model, Omnifact displays a Reasoning Effort picker directly in the footer of the model menu. This gives you precise control over how much deliberation the model applies to your prompt.

​
Adjusting the Effort Level

To adjust the thinking depth of an active reasoning model:
1

Select a Reasoning Model

Open the Model Selector in the chat input and pick a reasoning-capable model. The Reasoning Effort row appears at the bottom of the popover menu, beneath the model list.
2

Open the Effort Menu

Hover over or focus on the Reasoning Effort row to open the flyout submenu. On mobile devices and smaller screens, the effort options appear inline directly in the menu.
The model selector menu with Claude Opus 5.5 selected and the Reasoning Effort flyout open, showing the Low, Medium, High, and Extra High levels with a checkmark next to Medium
3

Choose an Effort Level

Click your preferred effort level. A checkmark confirms your active selection, and both the menu footer and the model selector in the chat input update to show the selected level (for example, “Claude Opus 5.5 Medium”).

​
Effort Levels Explained

Different tasks benefit from different amounts of deliberation. Omnifact supports up to four reasoning effort levels, depending on what the active model accepts:
Model-Specific Availability: Each AI model defines which effort levels it natively supports. For example, some models accept Low, Medium, and High, while others also support Extra High. The reasoning flyout automatically shows only the options available for your selected model. For non-reasoning models, the Reasoning Effort row is hidden entirely.

​
Per-Model Memory and Automatic Clamping

Omnifact manages your reasoning settings automatically across conversations:
  • Saved per Model: Your preferred effort level is saved in your browser for each model individually. When you switch back to a previously configured model, your preferred setting is restored.
  • Automatic Downgrade (Clamping): If you switch from a model set to Extra High to a model that only supports up to High, Omnifact automatically clamps your setting to the nearest supported level. If two supported levels are equidistant, Omnifact selects the lower, more cost-effective option.
  • Clean Interface: When switching back to a standard model, the reasoning controls disappear automatically, keeping your interface simple and focused.

​
Inspecting the Reasoning Trace

When a reasoning model responds in chat, you can review its thought process:
  1. Thinking Status: While the agent is working, an in-chat status indicator displays “Thinking…” to show that reasoning is actively underway.
  2. Expandable Reasoning Block: Above the final answer, Omnifact displays a collapsible Thinking card.
  3. Inspect the Logic: Click Show More to expand the step-by-step reasoning trace, or Show Less to collapse it. This allows you to verify the model’s intermediate deductions, assumptions, and calculations.
For strategies on structuring complex prompts that work hand-in-hand with reasoning models, see Thinking Step by Step.

​
Why Switch Models?

Different AI models are trained differently and have unique strengths. Switching models can help you optimize your results based on what you are trying to achieve. For a comprehensive guide on which model to choose for specific tasks, see the Model Guide.
Experiment! If you aren’t satisfied with a response, try switching to a different model or adjusting the reasoning effort, then click Regenerate on the last message to see how the AI approaches the same prompt with a fresh perspective.