Skip to main content
Cassidy gives you access to all leading AI models and automatically recommends the best options for different scenarios. You can choose models per Agent, per chat, and per Workflow step - so you always use the right tool for the job. For most work, start with Recommended Models. Browse the full list when you want a specific model, thinking level, or provider.

​
Where you choose models

Choose any foundational model as the base for your custom Agents. You can also use default Agents that run on foundational models without your business context.
Model selector in the Agent configuration panel
Cassidy’s Recommended Models section is continuously updated as new models are released and evaluated. The platform surfaces the best-performing, most cost-effective models for your needs, and cost reductions from providers are automatically passed on to you.
Recommended Models section showing Default, Fastest, Best Writer, and Smartest
You do not need to track AI model releases yourself - Cassidy keeps recommendations current as new models launch. If you are not sure which model to use, choose Default.

​
Model providers and strengths

Each provider has distinct strengths. Use this as a starting point, then experiment to find what works best for your specific tasks.

GPT (OpenAI)

Best for general Agent applications, everyday tasks, and Q&A. Reliable for information retrieval and a wide range of business queries. Excels at automation and general knowledge tasks.

Claude (Anthropic)

Preferred for creative writing, sales, and marketing content that needs a human-like touch. Ideal for drafting emails, marketing copy, and communication-heavy tasks that require natural-sounding language.

Gemini (Google)

Strong at combining information from many sources and working with documents, images, and other files. A good fit for deep research and analysis that pulls from a lot of material at once.

​
Which model and how much thinking

When you look past Recommended, two settings matter: which model you pick, and how much thinking it does. Not all models within a provider are the same. Some are faster and cheaper; others are more capable and cost more.

​
Thinking level

Thinking level controls how long the model works before it answers. Thinking variants use the same credit rate as Instant for that model. They cost more per task because they generate extra thinking tokens and run longer before answering. You can use this analogy to understand the model and thinking level choices: model power is how smart the brain is, and thinking level is how long that brain gets to work.

​
Understanding credit consumption

Credit usage scales with prompt size and response length. Longer prompts, more Knowledge Base context, and more thinking all use proportionally more credits.
Start with standard general-purpose models (or the Default recommended pick) since they offer excellent performance at the best value. Switch to High-powered or High Thinking models only when you need their specialized capabilities or the task needs complex reasoning.
For more on managing credits, see Manage credits.

​
Best practices

  • Start with Default unless you already know you need Fastest, Best Writer, or Smartest.
  • Use Mini models for summarization, formatting, and other high-volume Workflow steps.
  • Turn thinking up (Thinking or High Thinking) when accuracy matters more than speed.
  • Use High-powered models for complex analysis, planning, or Agents that rely on tools.
  • In a single Workflow, mix models: Mini for routine steps, Standard for generating content, High-powered or High Thinking only where the extra cost pays off.

​
Consider your specific use case

​
Next steps

Build and configure an Agent

Create a custom Agent and select the right model for its role.

Prompt engineering guide

Write effective prompts that get the most out of any model.