Chat mode and Simple generation
This feature is currently being tested and may not be visible to all users.
Chat mode with LLM
When the brain icon is on, Recraft uses conversational generation. In this mode:- The system keeps track of your history and context within the project.
- You can speak naturally, refine ideas, and ask for suggestions.
- The model may automatically adjust style, apply model choices for best results, or offer recommendations.
- You can continue evolving images step by step without re-describing everything.
- Edits and refinements follow the flow of the conversation.
- This workflow may consume slightly more credits and take longer per request, but may overall be a faster way to achieve desired results.
Simple generation mode
When the brain icon is off, Recraft switches to direct prompt-based generation:- No context or memory is used.
- The system does not treat the input as conversation, and each request is independent.
- You see faster generation and use fewer credits.
- The interface behaves like a streamlined image generator: type a prompt, get a result.
Switching between modes
You can toggle the brain icon at any time. Switching does not reset your work:- If you turn Chat mode on, the model can read the project’s existing history and continue from it.
- If you turn Chat mode off, you can continue refining images manually using the standard image generation model.
- You can move between modes as your workflow changes. For example, explore ideas with Chat mode on, then turn it off to fine-tune details more directly.
When to use each mode
| Mode | Best for | Behaviors |
|---|---|---|
| Chat mode | Guidance, exploration, automation | Remembers context, suggests improvements, may apply style/model automatically |
| Simple generation | Precision, manual control, simple workflow | No memory, no suggestions; strict prompt → image |