Flow Reasoning Models (FRMs) are a new framework that turns continuous flow models into efficient recurrent reasoners for structured tasks. By self‑conditioning a flow model on its own past outputs, FRMs iteratively refine solutions, allowing parallel decision making and revision. The authors introduce Fixed‑Point Forcing (FPF) to mitigate exposure bias at deeper recursion, and report near‑perfect solve rates on Sudoku‑Extreme, Zebra, and Maze‑Unique, outperforming existing masked‑diffusion and specialized baselines while using far fewer inference FLOPs.
By Alec Helbling, Andrey Bryutkin, Mauro Martino, Duen Horng Chau, Nima Dehmamy, Hendrik Strobelt
Flow matching enables language generation in few steps, but whether additional integration steps improve reasoning remains unclear. We prove that a flow parameterized by a two-layer Transformer can so...
arXiv:2610.09759v1 Announce Type: new
Abstract: Flow matching enables language generation in few steps, but whether additional integration steps improve reasoning remains unclear. We prove that a flo...
By Faissal Izermine, Hanru Bai, Oscar Davis, T. Konstantin Rusch
Hierarchical Continuous Diffusion Language Models (HC‑DLM) combine discrete token generation with a continuous latent trajectory in a single denoising process, addressing the independence bottleneck of parallel decoding in discrete diffusion models. The training objective is derived from a variational bound on token likelihood, and the latent state is the sole persistent generative element, with tokens read out and fed back at each step. Experiments on Sudoku, Countdown, and LM1B show HC‑DLM outperforming both discrete and continuous diffusion baselines in puzzle accuracy and generative perplexity.
By Hui Ren, Zihan Li, Chang Liu, Huidong Liu, Alexander Schwing
arXiv:2606. 27617v1 Announce Type: cross Abstract: Masked Diffusion Models (MDMs) promise fast, parallel language generation, but their reverse transition factorises across token positions -- an approximation that breaks down in the few-step sampling regime where parallel generation ought to provide the greatest efficiency gains.
By Iskander Azangulov, Kianoosh Ashouritaklimi, Leo Zhang, Simon Vary, Patrick Rebeschini
Draft Refinement Language Models (DRLMs) propose parallel revisions to a draft answer using a single denoiser that reconstructs every token from uniformly corrupted drafts, enabling both preservation and repair of tokens. The model evaluates each candidate draft by estimating a cavity energy from forward predictions, selecting revisions that lower this energy without a separate scoring model. Experiments on Countdown, Sudoku-Extreme, and Maze-Hard show DRLM outperforming autoregressive and discrete diffusion baselines, achieving 44.4% on Countdown-5 and 93.9% on Sudoku-Extreme with 6M parameters.
By Jaesang Yu, Seoyeon Byun, Yun Wang, Yan Lu, Juho Kim
Latent Recurrent Thoughts (LRT) proposes a method for reasoning with frozen large language models by operating in the model’s continuous representation space. A small auxiliary network generates initial latent vectors, which a tiny recurrent reasoner refines over multiple steps, decoupling computational depth from model size. Experiments on symbolic and natural‑language reasoning tasks show that LRT outperforms prior frozen‑decoder continuous‑space methods and chain‑of‑thought prompting while using far less inference compute.
By Zhaoliang Chen, Jie Fu
The paper introduces Stitching Noisy Diffusion Thoughts, a self‑consistency framework that transforms low‑cost diffusion‑sampled reasoning into a reusable pool of step‑level candidates. By sampling many diverse trajectories with a masked diffusion language model, scoring each intermediate step with a process reward model, and stitching the highest‑quality steps across trajectories, the method recomputes only the final answer, separating exploration from evaluation. Experiments on math reasoning benchmarks show that step‑level recombination improves accuracy by up to 23.8% on six tasks while reducing latency by up to 1.8× compared to traditional diffusion models and unified architectures.
By Roy Miles, Aysim Toker, Andreea-Maria Oncescu, Jiankang Deng, Ismail Elezi
arXiv:2607. 01170v1 Announce Type: cross Abstract: Generative reasoning re-rankers achieve strong recommendation accuracy by emitting a chain-of-thought before re-ordering a candidate list, but they are slow at inference: an autoregressive (AR) decoder spends one sequential forward pass per reasoning token, and the reasoning trace far exceeds the ranking it produces.
By Zhuoxuan Zhang (Yang), Kangqi Ni (Yang), Yuhang Chen (Yang), Mingfu Liang (Yang), Xiaohan Wei (Yang), Yunchen Pu (Yang), Fei Tian (Yang), Chonglin Sun (Yang), Frank Shyu (Yang), Adam (Yang), Song, Sandeep Pandey, Luke Simon, Tianlong Chen, Xi Liu
arXiv:2603. 01331v3 Announce Type: replace-cross Abstract: Discrete diffusion language models (dLLMs) generate text by iteratively denoising a masked sequence.
By Kejing Xia, Mingzhe Li, Lixuan Wei, Zhenbang Du, Xiangchi Yuan, Dachuan Shi, Qirui Jin, Wenke Lee
arXiv:2510. 19990v2 Announce Type: replace Abstract: The reasoning paradigm, where language models reason before answering, has enabled breakthroughs on tasks such as mathematical problem-solving.
By Zachary Horvitz, Raghav Singhal, Hao Zou, Carles Domingo-Enrich, Zhou Yu, Rajesh Ranganath, Kathleen McKeown
arXiv:2509. 21474v4 Announce Type: replace Abstract: While diffusion language models (DLMs) have achieved competitive performance in text generation, improving their reasoning ability with reinforcement learning remains an active research area.
By Guanghan Wang, Gilad Turok, Yair Schiff, Marianne Arriola, Volodymyr Kuleshov