Trending Model:#1Inklingthinkingmachines⬇16kTrending Model:#2Ternary-Bonsai-27B-ggufprism-ml⬇432kTrending Model:#3Unlimited-OCRbaidu⬇2237kTrending Model:#4Bonsai-27B-ggufprism-ml⬇1405kTrending Model:#5GLM-5.2zai-org⬇545kTrending Model:#6Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-NEO-MAX-MTP-GGUFDavidAU⬇63kTrending Model:#7Qwen3.6-35B-A3B-Uncensored-HauhauCS-AggressiveHauhauCS⬇1998kTrending Model:#8Qwythos-9B-Claude-Mythos-5-1M-GGUFempero-ai⬇2133kTrending Model:#9krea2-identity-editconradlocke⬇0kTrending Model:#10Laguna-S-2.1poolside⬇3kTrending Model:#1Inklingthinkingmachines⬇16kTrending Model:#2Ternary-Bonsai-27B-ggufprism-ml⬇432kTrending Model:#3Unlimited-OCRbaidu⬇2237kTrending Model:#4Bonsai-27B-ggufprism-ml⬇1405kTrending Model:#5GLM-5.2zai-org⬇545kTrending Model:#6Qwen3.6-27B-Fable-Fusion-711-Uncensored-Heretic-NM-DAU-NEO-MAX-MTP-GGUFDavidAU⬇63kTrending Model:#7Qwen3.6-35B-A3B-Uncensored-HauhauCS-AggressiveHauhauCS⬇1998kTrending Model:#8Qwythos-9B-Claude-Mythos-5-1M-GGUFempero-ai⬇2133kTrending Model:#9krea2-identity-editconradlocke⬇0kTrending Model:#10Laguna-S-2.1poolside⬇3k

Kezmark Deploys Mordant-12B-Think To Weave Detailed Image Prompts

By vramkickedin | | 2 min read
Close up of a fountain pen of high gradientscolors of dark crimson and contrasting to deep gold.

Mordant-12B-Think is a new AI model designed to generate detailed and narratively rich compositions for image generation. It uses chain-of-thought reasoning to analyze spatial relationships and construct deliberate meaning through compositional choices. This release focuses on providing high-quality prompt structures rather than just simple descriptions.

Developer Kezmark created this project by training it on a personally curated dataset of 7,203 highly detailed compositions. They added reverse-engineered chain-of-thought thinking bubbles to every training example to improve the analytical depth. The developer also gave the reasoning phase a sharp and caustic professional mentor personality that stays hidden in the final output.

Composition and reasoning tools

Key Features
  • Generates highly detailed structured image prompts.
  • Includes reverse-engineered chain-of-thought reasoning steps.
  • Trained on 7,203 highly detailed compositions.
  • Features a custom ComfyUI prompt enhancer.

This tool is ideal for creators who need to generate structured image prompts from short descriptions. Users can benefit from its ability to handle complex multi-element scenes requiring narrative-level spatial reasoning. It works best in situations where quality and detail matter more than generation speed.

Development and training notes

Kezmark expanded the training dataset from 3,804 examples to a total of 7,203 to improve the model. The developer decided to add a bit of snark and uniqueness to the reasoning phase to make the project feel more personal. This internal personality remains isolated and does not appear in the final composition output.

"And on top of that I decided to reverse-engineer CoT for all of them, because the hope was that would make the model better." - Source: Reddit

Project Links