mlsanityData to 7 Oct 2026

LLaDA2.1: Speeding Up Text Diffusion via Token Editing

arXiv 2602.08676 ↗First posted 9 Feb 2026cs.LG cs.AIcited by 48

On the map: Diffusion models → Discrete and text diffusion

Abstract (arXiv)

While LLaDA2.0 showcased the scaling potential of 100B-level block-diffusion models and their inherent parallelization, the delicate equilibrium between decoding speed and generation quality has remained an elusive frontier. Today, we unveil LLaDA2.1, a paradigm shift designed to transcend this trade-off. By seamlessly weaving Token-to-Token (T2T) editing into the conventional Mask-to-Token (M2T) scheme, we introduce a joint, configurable threshold-decoding scheme. This structural innovation gives rise to two distinct personas: the Speedy Mode (S Mode), which audaciously lowers the M2T threshold to bypass traditional constraints while relying on T2T to refine the output; and the Quality Mode (Q Mode), which leans into conservative thresholds to secure superior benchmark performances with manageable efficiency degrade. Furthering this evolution, underpinned by an expansive context window, we implement the first large-scale Reinforcement Learning (RL) framework specifically tailored for dLLMs, anchored by specialized techniques for stable gradient estimation. This alignment not only sharpens reasoning precision but also elevates instruction-following fidelity, bridging the chasm between diffusion dynamics and complex human intent. We culminate this work by releasing LLaDA2.1-Mini (16B) and LLaDA2.1-Flash (100B). Across 33 rigorous benchmarks, LLaDA2.1 delivers strong task performance and lightning-fast decoding speed. Despite its 100B volume, on coding tasks it attains an astounding 892 TPS on HumanEval+, 801 TPS on BigCodeBench, and 663 TPS on LiveCodeBench.

Key facts. Cited by 48 papers in the map (data through 7 Oct 2026). Placed in Discrete and text diffusion, one of 14 directions in Diffusion models. Builds on 40 of its 48 references that are in the map. First posted on arXiv 9 Feb 2026.

Builds on

40 of 48 references.

Closest ancestors · 7

Same direction, newest first

  1. 2025LLaDA2.0: Scaling Up Diffusion Language Models to 100B10 Dec
  2. 2025SPG: Sandwiched Policy Gradient for Masked Diffusion Language Models10 Oct
  3. 2025dInfer: An Efficient Inference Framework for Diffusion Language Models9 Oct
  4. 2025Revolutionizing Reinforcement Learning Framework for Diffusion Large Language Models8 Sep
  5. 2025Seed Diffusion: A Large-Scale Diffusion Language Model with High-Speed Inference4 Aug
  6. 2025Block Diffusion: Interpolating Between Autoregressive and Diffusion Language Models12 Mar
  7. 2025Remasking Discrete Diffusion Models with Inference-Time Scaling1 Mar

Other references · 33

  1. 2019HellaSwag: Can a Machine Really Finish Your Sentence?cited by 4,018
  2. 2017TriviaQA: A Large Scale Distantly Supervised Challenge Dataset for Reading Comprehensioncited by 2,895
  3. 2019PIQA: Reasoning about Physical Commonsense in Natural Languagecited by 2,818
  4. 2023GPQA: A Graduate-Level Google-Proof Q&A Benchmarkcited by 2,748
  5. 2024LiveCodeBench: Holistic and Contamination Free Evaluation of Large Language Models for Codecited by 1,906
  6. 2022Challenging BIG-Bench Tasks and Whether Chain-of-Thought Can Solve Themcited by 1,767
  7. 2024MMLU-Pro: A More Robust and Challenging Multi-Task Language Understanding Benchmarkcited by 1,758
  8. 2018Know What You Don't Know: Unanswerable Questions for SQuADcited by 1,538
  9. 2023Is Your Code Generated by ChatGPT Really Correct? Rigorous Evaluation of Large Language Models for Code Generationcited by 1,332
  10. 2023Instruction-Following Evaluation for Large Language Modelscited by 1,143
  11. 2018Spider: A Large-Scale Human-Labeled Dataset for Complex and Cross-Domain Semantic Parsing and Text-to-SQL Taskcited by 1,120
  12. 2024OlympiadBench: A Challenging Benchmark for Promoting AGI with Olympiad-Level Bilingual Multimodal Scientific Problemscited by 1,099
21 more
  1. 2019DROP: A Reading Comprehension Benchmark Requiring Discrete Reasoning Over Paragraphscited by 883
  2. 2023C-Eval: A Multi-Level Multi-Discipline Chinese Evaluation Suite for Foundation Modelscited by 694
  3. 2023Can LLM Already Serve as A Database Interface? A BIg Bench for Large-Scale Database Grounded Text-to-SQLscited by 667
  4. 2024BigCodeBench: Benchmarking Code Generation with Diverse Function Calls and Complex Instructionscited by 527
  5. 2022Language Models Are Greedy Reasoners: A Systematic Formal Analysis of Chain-of-Thoughtcited by 334
  6. 2024CRUXEval: A Benchmark for Code Reasoning, Understanding and Executioncited by 294
  7. 2024Omni-MATH: A Universal Olympiad Level Mathematic Benchmark For Large Language Modelscited by 277
  8. 2025AReaL: A Large-Scale Asynchronous Reinforcement Learning System for Language Reasoningcited by 234
  9. 2023MuSR: Testing the Limits of Chain-of-thought with Multistep Soft Reasoningcited by 195
  10. 2025ZebraLogic: On the Scaling Limits of LLMs for Logical Reasoningcited by 186
  11. 2024GSM-Plus: A Comprehensive Benchmark for Evaluating the Robustness of LLMs as Mathematical Problem Solverscited by 155
  12. 2025BIG-Bench Extra Hardcited by 123
  13. 2023CMATH: Can Your Language Model Pass Chinese Elementary School Math Test?cited by 89
  14. 2020OCNLI: Original Chinese Natural Language Inferencecited by 86
  15. 2024KOR-Bench: Benchmarking Language Models on Knowledge-Orthogonal Reasoning Taskscited by 67
  16. 2025PHYBench: Holistic Evaluation of Physical Perception and Reasoning in Large Language Modelscited by 57
  17. 2025Every Step Evolves: Scaling Reinforcement Learning for Trillion-Scale Thinking Modelcited by 50
  18. 2025ParallelBench: Understanding the Trade-offs of Parallel Decoding in Diffusion LLMscited by 42
  19. 2024ReaL: Efficient RLHF Training of Large Language Models with Parameter Reallocationcited by 34
  20. 2025Principled RL for Diffusion LLMs Emerges from a Sequence-Level Perspectivecited by 30
  21. 2025Effective Test-Time Scaling of Discrete Diffusion through Iterative Refinementcited by 7

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 48.

48 papers.

Sort:

By area:

  1. 2026Nemotron-Labs-Diffusion: A Tri-Mode Language Model Unifying Autoregressive, Diffusion, and Self-Speculation Decodingcited by 19
  2. 2026DMax: Aggressive Parallel Decoding for dLLMscited by 14
  3. 2026Learn from Your Mistakes: Self-Correcting Masked Diffusion Modelscited by 12
  4. 2026Introspective Diffusion Language Modelscited by 6
  5. 2026S2D2: Fast Decoding for Diffusion LLMs via Training-Free Self-Speculationcited by 5
  6. 2026DARE: Diffusion Large Language Models Alignment and Reinforcement Executorcited by 5
  7. 2026GDSD: Reinforcement Learning as Guided Denoiser Self-Distillation for Diffusion Language Modelscited by 5
  8. 2026DreamReasoner-8B: Block-Size Curriculum Learning for Diffusion Reasoning Modelscited by 4
  9. 2026ReflectDrive-2: Reinforcement-Learning-Aligned Self-Editing for Discrete Diffusion Drivingcited by 3
  10. 2026Which Tokens Matter? Adaptive Token Selection for RLVR with the Relative Surprisal Indexcited by 3
  11. 2026TAD: Temporal-Aware Trajectory Self-Distillation for Fast and Accurate Diffusion LLMcited by 2
  12. 2026TrajDLM: Topology-Aware Block Diffusion Language Model for Trajectory Generationcited by 2
36 more
  1. 2026Self-Generated Error Training for Token Editing in Diffusion Language Modelscited by 2
  2. 2026Multi-Block Diffusion Language Modelscited by 2
  3. 2026Don't Commit Alone: Joint Token Commitment in Diffusion Language Modelscited by 2
  4. 2026Rethinking the Generation Order of Block Diffusion Language Modelscited by 2
  5. 2026Generalized Discrete Diffusion with Self-Correctioncited by 1
  6. 2026Remask, Don't Replace: Token-to-Mask Refinement in Diffusion Language Modelscited by 1
  7. 2026dFlowGRPO: Rate-Aware Policy Optimization for Discrete Flow Modelscited by 1
  8. 2026D^2SD: Accelerating Speculative Decoding with Dual Diffusion Draft Modelscited by 1
  9. 2026Re-evaluating Confidence Remasking in Masked Diffusion Language Modelscited by 1
  10. 2026Efficient On-Device Diffusion LLM Inference with Mobile NPUcited by 1
  11. 2026Masked Diffusion Language Models are Strong and Steerable Text-Based World Models for Agentic RLcited by 1
  12. 2026FlowBlock: Wavefront-Parallel Decoding for Self-Correcting Diffusion Language Modelscited by 1
  13. 2026Beyond Block Boundaries: Multi-Block Editing for Diffusion Large Language Modelscited by 1
  14. 2026Speculative Correction: Draft-then-Refine Decoding for Diffusion Language Modelscited by 1
  15. 2026Block3D: Efficient Text-to-3D Generation via Block-Wise Diffusioncited by 1
  16. 2026GMTS: Gradient Magnitude-based Token Selection Improves RLVR Training for LLM Reasoningcited by 1
  17. 2026Unlocking Lossless Speedups in LLMs via Discrete Diffusioncited by 1
  18. 2026Controllable Accent Normalization via Discrete Diffusioncited by 0
  19. 2026MemDLM: Memory-Enhanced DLM Trainingcited by 0
  20. 2026CARE: Privacy-Compliant Agentic Reasoning with Evidence Discordancecited by 0
  21. 2026Block-R1: Rethinking the Role of Block Size in Multi-domain Reinforcement Learning for Diffusion Large Language Modelscited by 0
  22. 2026Extracting Training Data from Diffusion Language Models via Infillingcited by 0
  23. 2026D2-Monitor: Dynamic Safety Monitoring for Diffusion LLMs via Hesitation Signalscited by 0
  24. 2026Targeted Remasking: Replacing Token Editing with Token-to-Mask Refinement in Discrete Diffusion Language Modelscited by 0
  25. 2026Decoding in Order-Agnostic Language Models: Chain-Rule Deviation and Uniform Spreadingcited by 0
  26. 2026Mean-Field Parallel Decoding for Discrete Diffusion Language Modelscited by 0
  27. 2026Nemotron-Labs-Diffusion-Image: Advancing Masked Discrete Diffusion for High-Resolution Image Synthesiscited by 0
  28. 2026From Interface to Inference: Eliciting Any-Order Inference from Any-Order Modelscited by 0
  29. 2026Retrofitting Linear Attention into Diffusion Language Modelscited by 0
  30. 2026A Ticket from Marginals to Joints: Coupled-Noise Distillation for One-Step Block Generation in Diffusion Language Modelscited by 0
  31. 2026DA-DLM: Explicitly Modeling Token Dependencies in Diffusion Language Modelscited by 0
  32. 2026OLED-MoE: Accelerating MoE-Based dLLM Inference via Inter-Iteration Locality-Aware Expert Offloadingcited by 0
  33. 2026From Position Risks to Block Survival: Faster Generation for Diffusion Language Modelscited by 0
  34. 2026Faster Block-Diffusion Serving with Distribution-Free Risk Guaranteescited by 0
  35. 2026DEdit: Iterative Draft Editing for Speculative Decodingcited by 0
  36. 2026SpecFold: Folding Multi-Branch Redundancy for Faster Speculative Decoding in Diffusion Language Modelscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Fast-dLLM: Training-free Acceleration of Diffusion LLM by Enabling KV Cache and Parallel Decoding
  2. 2026Just on Time: Token-Level Early Stopping for Diffusion Language Models
  3. 2026Locally Confident, Globally Stuck: The Quality-Exploration Dilemma in Diffusion Language Models
  4. 2026Efficient Diffusion LLMs via Temporal-Spatial Parallel Decoding and Confidence Extrapolation
  5. 2025Beyond Next-Token Prediction: A Performance Characterization of Diffusion versus Autoregressive Language Models
  6. 2026The Path Matters: Learning a Token-Commitment Policy for Diffusion Language Models
  7. 2026Diffusion-State Policy Optimization for Masked Diffusion Language Models
  8. 2025Don't Settle Too Early: Self-Reflective Remasking for Diffusion Language Models