Skip to content

Choosing a model

The agent is powered by a large language model, and you choose which one — per project. Different models have different strengths, so you can match the model to the work.

Where to choose

Open the model picker in the workspace and select the model for the current project. The change applies to the agent's next turn, so you can switch mid-project if you want a different model for a particular task.

Available providers

Factorly supports models across three providers:

  • Claude — via Amazon Bedrock. Anthropic's models, served through AWS Bedrock.
  • OpenAI. GPT-family models.
  • OpenRouter. A gateway that opens up a wide catalog of additional models.

Choosing well

There's no single "best" model — it depends on the job:

  • For complex, multi-file features and careful refactors, reach for a stronger reasoning model.
  • For quick edits and iteration, a faster model keeps the loop snappy and costs down (see credits & billing).

A good habit: build the hard parts with a top-tier model, then switch to a faster one for polish and small tweaks.

Next: put your app on the internet with Deploying.