The Generate Text step lets you prompt AI models. You can use any of the models we have available on Respell, with most of their settings available.

Generate Text step


ModelDropdownChoose the AI model you want a response from.
PromptPlain TextThe prompt (instruction) to send to the AI. This could be a question or a task you give the AI to complete.
Advanced ModeBooleanEnable the advanced options for this step. The options below are hidden unless this is enabled.
TemperatureDecimal between 0-2How creative the AI should be in its response. Higher temperatures will make the AI more creative, but it may not follow the prompt entirely. Set temperature low for tasks where you want consistent responses.
Max TokensInteger between 0-4096Maximum number of tokens to generate. The allowed values will be different depending on the selected model.
JSON ModeBooleanWhether the output should be returned in JSON format.
System InstructionPlain TextDefine how the model should behave when writing responses. This is where you tell the model how it should sound, who to roleplay as, or other instructions that dictates its identity.


ResponsePlain TextThe response from the AI model.


  • We have curated steps (such as Summarize Text) with pre-built prompts for common steps. Try those steps before writing your own prompt!
  • Try different models with your prompt to see how it performs. Different models are good at some tasks, but may struggle with others.
  • Set temperature to 1 to start while you’re experimenting.
  • A lower Max Tokens setting will make the model run faster.
  • When writing your own prompt, you’ll need to move between editing the prompt and testing a lot. We’re working on an experience to help with this! 🤫