# Oddin: Latent Visual Reasoning in Unified Multimodal Models via Self-Distillation from Image Generation

Authors: Jaeyong Ko, Hankyeol Kim, Pilsung Kang.

Replaces intermediate generated images with learned latent-token blocks while retaining the written reasoning between them. Self-distills successful text-image traces into a unified multimodal model, reserving image generation for the final output.

First public: 2026-10-19.
ArXiv initial submission: 2026-10-19.

[Original paper](https://arxiv.org/abs/2610.33062) · [PDF](https://arxiv.org/pdf/2610.33062) · [HTML](https://arxiv.org/html/2610.33062v1) · [Reading-list entry](https://latent-implicit-thinking.github.io/papers/2610.33062/)
[Preprint BibTeX](https://latent-implicit-thinking.github.io/papers/2610.33062/citation.bib) · [Preprint CSL-JSON](https://latent-implicit-thinking.github.io/papers/2610.33062/citation.csl.json) · [JSON record](https://latent-implicit-thinking.github.io/papers/2610.33062/record.json)

## Approaches

- [Thinking tokens](https://latent-implicit-thinking.github.io/topics/thinking-tokens/): Give a model extra tokens for computation, such as pauses, filler symbols, or learned planning tokens. These tokens need not spell out the solution.
- [Continuous-state reasoning](https://latent-implicit-thinking.github.io/topics/continuous-states/): Use vectors—lists of numbers—to carry intermediate reasoning from one step to the next, instead of selecting a text token at every step. Some methods use the model's hidden states; others combine several candidate token representations.

Reading-list version 0.12.0; literature checked through 2026-11-02.