Character.AI Launches Image Models for Consistent Fan Stories

Character.AI has put a new family of image models into its app, aiming to make fan stories visually coherent rather than merely attractive. CAI-Image powers (c.ai) Comics, Imagine and Imagine Message, with the company emphasizing recognizable Characters across changing scenes, styles and comic pages.

Character.AI describes CAI-Image as a set of post-trained versions of the open-source Qwen-Image model. The company says it began with the demands of its own character-led creation flow: users can generate images from chats and premises, where the system assembles prompts and brings the relevant Characters into a scene.

That framing explains the release’s focus. Character.AI says the models were tuned for style transfer, multi-Character scenes and pose control, camera control, and manga and comics generation. It also says a generation can combine Character images with pose, style and location references, rather than asking users to resolve those inputs separately.

245Natural-language camera positions

Character.AI says CAI-Image understands 245 camera positions, organized across five distances and seven horizontal and seven vertical angles.

A single comic page concentrates the problem: panel structure, dialogue, emotion, camera variation and visual continuity all need to work together. Character.AI has included a dedicated manga model that it says is trained for those pages, including accurate dialogue text and a cast that remains recognizable from panel to panel.

The company says the models run on CAI-MM-Studio, its multimodal system for data collection and cleaning, multi-node training, automated evaluation and serving. Character.AI argues that controlling those stages lets it tailor models to its app’s workflow and operate them quickly and cost-effectively; those speed and cost advantages are company claims, not independently reported comparisons.

Character.AI’s stated next step is CAI-Image V2, which it says will work on consistent environments as the camera moves, more nuanced emotion between Characters, and better manga layouts and text. That is a useful boundary on today’s release: the company is presenting Character continuity as the foundation, while scene continuity and more reliable comic construction remain work in progress.