Which AI model powers Communication Designer?
All generators run on Text and image models are served through Lovable AI Gateway; exact models vary by tool.. Text and strategy tools use the flash and pro variants, image tools use flash-image.
There is one gateway in front of the models, so a tool can be routed to a different variant when a task needs more reasoning depth (for example a design brief) or faster turnaround (for example a meme caption).
Video, GIF and motion output is not a native video model: keyframes are generated as images and assembled into a WebM clip in your browser. That is why those clips have no audio track.
Trend research and competitor analysis do not run a live web crawl of the whole internet. Trend research works from model knowledge; competitor analysis fetches the page text of a URL you supply and cannot read login-protected pages or execute their JavaScript.
事实
- Models: Text and image models are served through Lovable AI Gateway; exact models vary by tool.
- Video/GIF/motion: browser-rendered keyframe clips, no audio
- Competitor analysis: reads fetched page text only
- Trend research: model knowledge, no source URLs