# Beyond Next-Token Prediction: A Performance Characterization of Diffusion versus Autoregressive Language Models

arXiv: https://arxiv.org/abs/2510.04146 · HTML: https://mlsanity.com/paper/2510.04146/

## Key facts

- Cited by 13 papers in the map (data through 14 Oct 2026).
- Builds on 30 of its 34 references that are in the map.
- First posted on arXiv 5 Oct 2025.

## Builds on (most cited first, 30 of 30)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [Denoising Diffusion Probabilistic Models](https://mlsanity.com/paper/2006.11239/) (2020)
- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [The Llama 3 Herd of Models](https://mlsanity.com/paper/2407.21783/) (2024)
- [Training Verifiers to Solve Math Word Problems](https://mlsanity.com/paper/2110.14168/) (2021)
- [Evaluating Large Language Models Trained on Code](https://mlsanity.com/paper/2107.03374/) (2021)
- [Hierarchical Text-Conditional Image Generation with CLIP Latents](https://mlsanity.com/paper/2204.06125/) (2022)
- [OPT: Open Pre-trained Transformer Language Models](https://mlsanity.com/paper/2205.01068/) (2022)
- [Fast Inference from Transformers via Speculative Decoding](https://mlsanity.com/paper/2211.17192/) (2022)
- [Diffusion-LM Improves Controllable Text Generation](https://mlsanity.com/paper/2205.14217/) (2022)
- [Accelerating Large Language Model Decoding with Speculative Sampling](https://mlsanity.com/paper/2302.01318/) (2023)
- [Large Language Diffusion Models](https://mlsanity.com/paper/2502.09992/) (2025)
- [WaveGrad: Estimating Gradients for Waveform Generation](https://mlsanity.com/paper/2009.00713/) (2020)
- [KVQuant: Towards 10 Million Context Length LLM Inference with KV Cache Quantization](https://mlsanity.com/paper/2401.18079/) (2024)
- [Structure and Content-Guided Video Synthesis with Diffusion Models](https://mlsanity.com/paper/2302.03011/) (2023)
- [Efficiently Scaling Transformer Inference](https://mlsanity.com/paper/2211.05102/) (2022)
- [DiffuSeq: Sequence to Sequence Text Generation with Diffusion Models](https://mlsanity.com/paper/2210.08933/) (2022)
- [Fast-dLLM: Training-free Acceleration of Diffusion LLM by Enabling KV Cache and Parallel Decoding](https://mlsanity.com/paper/2505.22618/) (2025)
- [SqueezeLLM: Dense-and-Sparse Quantization](https://mlsanity.com/paper/2306.07629/) (2023)
- [Mercury: Ultra-Fast Language Models Based on Diffusion](https://mlsanity.com/paper/2506.17298/) (2025)
- [Speculative Decoding with Big Little Decoder](https://mlsanity.com/paper/2302.07863/) (2023)
- [dKV-Cache: The Cache for Diffusion Language Models](https://mlsanity.com/paper/2505.15781/) (2025)
- [Self-conditioned Embedding Diffusion for Text Generation](https://mlsanity.com/paper/2211.04236/) (2022)
- [Full Stack Optimization of Transformer Inference: a Survey](https://mlsanity.com/paper/2302.14017/) (2023)
- [FlashDLM: Accelerating Diffusion Language Model Inference via Efficient KV Caching and Guided Diffusion](https://mlsanity.com/paper/2505.21467/) (2025)
- [Speculative Diffusion Decoding: Accelerating Language Generation through Diffusion](https://mlsanity.com/paper/2408.05636/) (2024)
- [Empowering Diffusion Models on the Embedding Space for Text Generation](https://mlsanity.com/paper/2212.09412/) (2022)
- [QuantSpec: Self-Speculative Decoding with Hierarchical Quantized KV Cache](https://mlsanity.com/paper/2502.10424/) (2025)
- [DLM-One: Diffusion Language Models for One-Step Sequence Generation](https://mlsanity.com/paper/2506.00290/) (2025)

## Built on by (most cited first, 13 of 13)

- [ParallelBench: Understanding the Trade-offs of Parallel Decoding in Diffusion LLMs](https://mlsanity.com/paper/2510.04767/) (2025)
- [CDLM: Consistency Diffusion Language Models For Faster Sampling](https://mlsanity.com/paper/2511.19269/) (2025)
- [How Efficient Are Diffusion Language Models? A Critical Examination of Efficiency Evaluation Practices](https://mlsanity.com/paper/2510.18480/) (2025)
- [Residual Context Diffusion Language Models](https://mlsanity.com/paper/2601.22954/) (2026)
- [MMaDA-VLA: Large Diffusion Vision-Language-Action Model with Unified Multi-Modal Instruction and Generation](https://mlsanity.com/paper/2603.25406/) (2026)
- [EfficientRollout: System-Aware Self-Speculative Decoding for RL Rollouts](https://mlsanity.com/paper/2606.18967/) (2026)
- [Pythagoras-Prover: Advancing Efficient Formal Proving via Augmented Lean Formalisation](https://mlsanity.com/paper/2606.12594/) (2026)
- [MAGE: All-[MASK] Block Already Knows Where to Look in Block Diffusion LLM](https://mlsanity.com/paper/2602.14209/) (2026)
- [Enabling KV Caching of Shared Prefix for Diffusion Language Models](https://mlsanity.com/paper/2606.07571/) (2026)
- [OLED-MoE: Accelerating MoE-Based dLLM Inference via Inter-Iteration Locality-Aware Expert Offloading](https://mlsanity.com/paper/2609.33385/) (2026)
- [d-OPD: Future-Aware On-Policy Distillation for Block Diffusion Language Models](https://mlsanity.com/paper/2609.35362/) (2026)
- [FastGuide: Accelerating Reward Guidance for Diffusion Large Language Models](https://mlsanity.com/paper/2609.36202/) (2026)
- [Plan-and-Patch: Diffusion Language Models for Agentic Planning](https://mlsanity.com/paper/2610.10786/) (2026)

## Abstract (arXiv)

Large Language Models (LLMs) have achieved state-of-the-art performance on a broad range of Natural Language Processing (NLP) tasks, including document processing and code generation. Autoregressive Language Models (ARMs), which generate tokens sequentially conditioned on all previous tokens, have been the predominant paradigm for LLMs. While these models have achieved high accuracy across a range of downstream tasks, they exhibit low arithmetic intensity due to the inherent sequential dependency in next-token prediction. Recently, Diffusion Language Models (DLMs) have emerged as a promising alternative architecture. DLMs generate output tokens in parallel, mitigating the limitations of sequential decoding. However, the performance implications of DLMs relative to commonly deployed ARMs are not fully understood. In this work, we present a comprehensive study of the performance characteristics of ARMs and DLMs, combining theoretical analysis with empirical profiling to characterize the trade-offs between these approaches. We show that although DLMs can achieve higher arithmetic intensity than ARMs by leveraging parallelism across token positions, they fail to scale effectively with longer contexts. We then explore block-wise decoding for DLMs, which decouples arithmetic intensity from sequence length and enables better scaling to long contexts (similar to ARMs). We also examine batched inference and find that ARMs exhibit superior throughput as they benefit more from parallelism across sequences in the batch. Finally, we highlight opportunities for accelerating DLM inference, emphasizing that reducing the number of sampling steps is key for open-source DLMs to achieve lower latency relative to ARMs.