method · first seen 4 Oct 2026
speculative decoding
Also: SD, speculative drafting, parallel action decoding
Inference acceleration where a drafter proposes multiple tokens that the target model verifies in parallel.
Speculative decoding speeds up autoregressive generation by having a cheap drafter propose several tokens that the target model checks in one forward pass, keeping the output distribution unchanged. The idea dates to 2022-2023 and is now standard in serving stacks. The sources show ongoing refinement rather than a new idea: papers adapt it to diffusion language models, looped transformers, long-context attention dilution, tree verification scheduling, session-local drafters, streaming speech recognition and robot action decoding. On the code side, an official EAGLE-4 release and several local inference engines for CPUs and Apple Silicon list it as a supported feature.
It is already a default technique for cutting LLM latency, so the activity here reflects how widely it is used rather than early adoption. The most interesting threads are its extension to non-autoregressive and looped architectures and to domains outside text generation.
Analyst note (established), 17 Nov 2026.
Mentions per week
PapersCodeDiscussion
Rank 36 on 14 Nov 2026 · 120 mentions in 28 days · 118 independent groups · papers → code → discussion · first seen 6 weeks ago · acceleration -0.24
Earliest sources
- SpecFold: Folding Multi-Branch Redundancy for Faster Speculative Decoding in Diffusion Language ModelsPapers · 4 Oct 2026 · Chung-En Ho, Weiyu Sun, Cheng-Jhih Shih, He Li, Yong Liu, Yingyan Celine Lin
- SharpDraft: Accelerating Long-Context Speculative Decoding with Cardinality-Aware Query ScalingPapers · 4 Oct 2026 · Jeonghoon Park, Seongwoon Jo, Jongwon Lee, Taesik Gong
- AdaSpark: Adaptive DSpark with Online Learning for Tree Verification and N-gram FillPapers · 5 Oct 2026 · Liquan Liu, Yifan Zhang, Bowei Xu
- Byte Language Models: Scaling, Emergent Abstractions, and Information AllocationPapers · 5 Oct 2026 · Jie Wang, Shiwei Luo, Qi Zhang, Yuanbin Wu
- D-Loop: Looped Diffusion Drafting for Speculative DecodingPapers · 5 Oct 2026 · Kecheng Chen, Yuyang He, Cheng Gong, Hui Liu, Guoping Long, Jiajun Li, Shi Wu, S…
- SchemaFill: Efficient LLM Tool Calling via Slot-Parallel Speculative DecodingPapers · 5 Oct 2026 · Zhi-Kai Chen, Song-Yan Li, De-Chuan Zhan, Han-Jia Ye
Latest sources
- sybil-solutions/inference-optimisation-skillCode · 16 Nov 2026 · sybil-solutions · 915 stars
- K2-Horizon: A Frontier Fleet of Open Source Language ModelsPapers · 16 Nov 2026 · K2 Team, Zhengzhong Liu, Haonan Li, Xuezhe Ma, Alexander Moreno, Subham Sekhar S…
- Nemotron 3.5 Bolt E4B 4GB: An Open and Efficient Small Language Model for Fast Conversational AgentsPapers · 16 Nov 2026 · NVIDIA, :, Adi Margolin, Aditya Malte, Aishwarya Chhabra, Alex Hodson, Ameya Sun…
- Memory-Efficient Speculative Decoding with Quantized Draft KV CachePapers · 16 Nov 2026 · Saksham Gupta, Michael Li
- Decode Early, Verify Late: Streaming Offline Speech Recognizers Without Accuracy LossPapers · 15 Nov 2026 · Doyeob Kim
- LoopSkim: Lossless Multi-Depth Speculative Decoding for Looped Language ModelsPapers · 15 Nov 2026 · Muhan Xiao, Shijian Li
- LoopFill: Pipelined Multi-Token Speculative Decoding for Looped TransformersPapers · 14 Nov 2026 · Siqi Zhu, Suozhi Huang, Yi Zhao, Jiaxuan You
- Text Speeding Itself Up: Recency and Restraint in Session-Local DraftingPapers · 14 Nov 2026 · Antal van den Bosch
- VLN-Flash: Parallel Action Decoding with Adaptive Execution for Vision-and-Language NavigationPapers · 14 Nov 2026 · Ruizhe Xie, Honghua Chen, Binglin Wang, Tong Guo, Xu Tang, Puti Yan, Shaoxiong C…
- DynaDFlash: Dynamic Feature Projection for Block Diffusion SpeculatorsPapers · 14 Nov 2026 · Aadi Kulshrestha, Reese Chong
- DeepSeek UltrafastDiscussion · 13 Nov 2026 · neo2006 · 71 points
- SpecRelay: Amortize or Avoid the Round Trip in Edge-Cloud Speculative DecodingPapers · 13 Nov 2026 · Jai Jain, Chihyo Ahn, Yong Liu, Charles Chen, Duc N. M Hoang, Ajay Kumar Jaiswal…