# Utility-Driven Speculative Decoding for Mixture-of-Experts

arXiv: https://arxiv.org/abs/2506.20675 · HTML: https://mlsanity.com/paper/2506.20675/

## Key facts

- Cited by 13 papers in the map (data through 14 Oct 2026).
- Placed in Efficient MoE training and inference systems, one of 12 directions in Mixture-of-experts.
- Builds on 29 of its 45 references that are in the map.
- First posted on arXiv 17 Jun 2025.

## Placed in

- Mixture-of-experts / [Efficient MoE training and inference systems](https://mlsanity.com/moe/efficient-moe-training-and-inference-systems/)
- Efficient LLM inference and serving / [Diffusion LLM and MoE inference](https://mlsanity.com/inference/diffusion-llm-and-moe-inference/)

## Builds on (most cited first, 29 of 29)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [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)
- [Judging LLM-as-a-Judge with MT-Bench and Chatbot Arena](https://mlsanity.com/paper/2306.05685/) (2023)
- [Evaluating Large Language Models Trained on Code](https://mlsanity.com/paper/2107.03374/) (2021)
- [Efficient Memory Management for Large Language Model Serving with PagedAttention](https://mlsanity.com/paper/2309.06180/) (2023)
- [Outrageously Large Neural Networks: The Sparsely-Gated Mixture-of-Experts Layer](https://mlsanity.com/paper/1701.06538/) (2017)
- [Switch Transformers: Scaling to Trillion Parameter Models with Simple and Efficient Sparsity](https://mlsanity.com/paper/2101.03961/) (2021)
- [Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone](https://mlsanity.com/paper/2404.14219/) (2024)
- [Mixtral of Experts](https://mlsanity.com/paper/2401.04088/) (2024)
- [Fast Inference from Transformers via Speculative Decoding](https://mlsanity.com/paper/2211.17192/) (2022)
- [No Language Left Behind: Scaling Human-Centered Machine Translation](https://mlsanity.com/paper/2207.04672/) (2022)
- [Accelerating Large Language Model Decoding with Speculative Sampling](https://mlsanity.com/paper/2302.01318/) (2023)
- [DeepSeekMoE: Towards Ultimate Expert Specialization in Mixture-of-Experts Language Models](https://mlsanity.com/paper/2401.06066/) (2024)
- [EAGLE: Speculative Sampling Requires Rethinking Feature Uncertainty](https://mlsanity.com/paper/2401.15077/) (2024)
- [EAGLE-2: Faster Inference of Language Models with Dynamic Draft Trees](https://mlsanity.com/paper/2406.16858/) (2024)
- [EAGLE-3: Scaling up Inference Acceleration of Large Language Models via Training-Time Test](https://mlsanity.com/paper/2503.01840/) (2025)
- [OLMoE: Open Mixture-of-Experts Language Models](https://mlsanity.com/paper/2409.02060/) (2024)
- [Atom: Low-bit Quantization for Efficient and Accurate LLM Serving](https://mlsanity.com/paper/2310.19102/) (2023)
- [Accelerating LLM Inference with Staged Speculative Decoding](https://mlsanity.com/paper/2308.04623/) (2023)
- [Pre-gated MoE: An Algorithm-System Co-Design for Fast and Scalable Mixture-of-Expert Inference](https://mlsanity.com/paper/2308.12066/) (2023)
- [PEARL: Parallel Speculative Decoding with Adaptive Draft Length](https://mlsanity.com/paper/2408.11850/) (2024)
- [MoE-Lightning: High-Throughput MoE Inference on Memory-constrained GPUs](https://mlsanity.com/paper/2411.11217/) (2024)
- [Towards MoE Deployment: Mitigating Inefficiencies in Mixture-of-Expert (MoE) Inference](https://mlsanity.com/paper/2303.06182/) (2023)
- [TurboSpec: Closed-loop Speculation Control System for Optimizing LLM Serving Goodput](https://mlsanity.com/paper/2406.14066/) (2024)
- [Ouroboros: Generating Longer Drafts Phrase by Phrase for Faster Speculative Decoding](https://mlsanity.com/paper/2402.13720/) (2024)
- [AdaSpec: Adaptive Speculative Decoding for Fast, SLO-Aware Large Language Model Serving](https://mlsanity.com/paper/2503.05096/) (2025)
- [Speculative Decoding with CTC-based Draft Model for LLM Inference Acceleration](https://mlsanity.com/paper/2412.00061/) (2024)
- [Mixture of Cache-Conditional Experts for Efficient Mobile Device Inference](https://mlsanity.com/paper/2412.00099/) (2024)

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

- [MoE-Spec: Expert Budgeting for Efficient Speculative Decoding](https://mlsanity.com/paper/2602.16052/) (2026)
- [Making Every Verified Token Count: Adaptive Verification for MoE Speculative Decoding](https://mlsanity.com/paper/2605.00342/) (2026)
- [DraftExpert: Expansion-Aware Self-Speculative Decoding for End-Device MoE Inference](https://mlsanity.com/paper/2607.24434/) (2026)
- [S2-MoE: Enabling Efficient Self-Speculative Decoding for Mixture-of-Experts on Edge Devices](https://mlsanity.com/paper/2608.15018/) (2026)
- [AcceptMoE: Commitment-Weighted Self-Sizing Verifier Expert Sets for Efficient MoE Speculative Decoding](https://mlsanity.com/paper/2608.02989/) (2026)
- [The Limits of Speculation: Bounding Speculative Decoding in Mixture-of-Experts](https://mlsanity.com/paper/2609.22156/) (2026)
- [ELMoE-3D: Leveraging Intrinsic Elasticity of MoE for Hybrid-Bonding-Enabled Self-Speculative Decoding in On-Premises Serving](https://mlsanity.com/paper/2604.14626/) (2026)
- [Rethinking Network Topologies for Cost-Effective Mixture-of-Experts LLM Serving](https://mlsanity.com/paper/2605.00254/) (2026)
- [WhiteMatter: All-to-All Cross-Layer Connections via KV Source Mixing](https://mlsanity.com/paper/2608.18486/) (2026)
- [BigMoMo: Efficient Inference of Large-Scale MoE with Speculative Decoding on Mobile Devices](https://mlsanity.com/paper/2609.14643/) (2026)
- [Efficient Mixture-of-Experts with Speculative Decoding via Expert Coactivation](https://mlsanity.com/paper/2609.22471/) (2026)
- [CAST: Cost-Aware Speculative Trees from One-Pass Block Drafters](https://mlsanity.com/paper/2610.00321/) (2026)
- [AdaSpark: Adaptive DSpark with Online Learning for Tree Verification and N-gram Fill](https://mlsanity.com/paper/2610.05774/) (2026)

## Abstract (arXiv)

GPU memory bandwidth is the main bottleneck for low-latency Large Language Model (LLM) inference. Speculative decoding leverages idle GPU compute by using a lightweight drafter to propose K tokens, which the LLM verifies in parallel, boosting token throughput. In conventional dense LLMs, all model weights are fetched each iteration, so speculation adds no latency overhead. Emerging Mixture of Experts (MoE) models activate only a subset of weights per token, greatly reducing data movement. However, we show that speculation is ineffective for MoEs: draft tokens collectively activate more weights, increasing data movement and verification time by 2-3x. When token throughput gains fail to offset this overhead, speculation causes slowdowns up to 1.5x, making it infeasible. Even when useful, the optimal K varies by task, model, and even between requests and iterations. Thus, despite widespread use in dense LLMs, speculation remains impractical in leading MoEs. We present Cascade, a utility-driven framework that selectively enables speculation to avoid slowdowns and dynamically tunes K to accelerate MoE serving. Cascade uses a lightweight metric, speculation utility, the ratio of token gains to verification cost, which shows iteration-level locality, enabling periodic decisions via short test and longer set phases. For each request, Cascade disables speculation if utility drops below one during testing, and when utility exceeds one, tests multiple K-values to choose the utility-maximizing K for the set phase. We implement Cascade in vLLM and evaluate it on five popular MoEs with workloads spanning code, math, extraction, and mixed tasks. Cascade limits slowdown to 5% (vs. 1.5x) and improves throughput by 7-14% over static K, making speculative decoding practical for MoEs.