Choosing an AI Model
Pick the model that fits your agent's needs - and your budget.
Written by Outlearn Documentation
Updated at April 27th, 2026
Table of Contents
The model your agent runs on shapes how it handles every conversation. Choose the right one and your agent reasons clearly, stays on point, and handles nuanced questions well. Choose the wrong one and you will notice it in the responses.
For most teams, the default model handles everyday support questions well - and you will not need to think about this setting at all. But if your users are technical, your questions are nuanced, or your conversations require a higher level of reasoning, choosing the right model is the difference between an agent that answers well and one that answers precisely.
Outlearn gives you a range of models to choose from, each with different strengths and credit costs. This article helps you understand what each one is good at so you can make a confident choice for your use case - and know when it is worth upgrading.
In this article, you'll learn:
- What an AI model is and why it matters
- What models are available
- How to choose the right one for your use case
- How to change your model at any time
What Is an AI Model?
Think of the AI model as the brain behind your agent. It's what reads your user's question, understands the context, searches your sources, and writes the response.
Different models have been trained differently - some are faster and lighter, some are more capable at handling complex or nuanced questions. The trade-off is usually capability vs. credit cost.
All available models are from OpenAI. Each one has a credit cost per conversation.
GPT 5 Mini is the default model. It offers strong performance at a low credit cost and is a good starting point for most teams.
How to Change Your AI Model
Outlearn supports multiple AI providers - OpenAI (GPT), Anthropic (Claude), and Google (Gemini).
You don’t need to overthink this. The right choice comes down to what matters most for your use case: quality, speed, or cost.
You can change your model at any time - the change takes effect immediately for all new conversations.
- Go to Settings.
- Under AI Model & Instructions, click the AI Model dropdown.
- Select the model you want to use.
- Click Save Changes.
Changing the model does not affect past conversations - only new ones going forward.
Choosing Between OpenAI, Claude, and Gemini Models
Claude (Best for clarity and deeper answers)
Claude models tend to give more structured, thoughtful responses and handle complex questions well.
Use Claude if:
- Your users ask multi-step or nuanced questions
- You want clearer, more polished answers
- Accuracy and explanation quality matter more than speed
Good starting points:
- Claude Sonnet (4.5 / 4.6) → best balance for most teams
- Claude Opus → for very complex or high-stakes queries
- Claude Haiku → faster, lighter option for simpler use cases
Gemini (Best for speed and scale)
Gemini models are fast and very cost-efficient, which makes them a strong choice for high-volume support.
Use Gemini if:
- You have a lot of conversations happening at once
- Questions are mostly straightforward
- You want to keep credit usage low
Good starting points:
- Gemini 2.5 Flash / Lite → fast and cost-efficient
- Gemini 2.5 Pro → better quality when needed
- Gemini 3 Pro (Preview) → for testing newer capabilities
OpenAI (Best all-around default)
OpenAI models are still the most balanced option across most use cases.
Use OpenAI if:
- You want a reliable default that performs well across the board
- Your use case is mixed (some simple, some complex questions)
Good starting points:
- GPT 5 Mini → default, great performance for cost
- GPT 4.1 / 4o → stronger reasoning when needed
If responses aren’t great, don’t switch models right away.
In most cases, improving your content and instructions will have a bigger impact than changing the model.
Best Practices
- Don't switch models just because responses aren't great - check your instructions and sources first. A well-configured agent on a lighter model will outperform a poorly configured agent on a powerful one.
- If you switch to a higher-credit model, keep an eye on your credit usage in the Billing section for the first week to understand the impact.
- Use the Preview panel to test model changes before they go live - ask it your five most common support questions and compare the quality of responses.