LLM Inference Optimization: Quantization, Speculative Decoding, and Beyond

Why Inference Optimization Matters

Training a large language model is a one-time cost. Inference — serving predictions to users — is the ongoing expense that determines whether a model is economically viable in production. A model that costs $10 million to train but $0.001 per query can generate billions of responses profitably. The same model at $0.10 per query may be commercially unviable.

Inference optimization is the discipline of making models faster, cheaper, and more memory-efficient without sacrificing output quality. Here are the techniques that matter most in 2026.

Quantization: Trading Precision for Speed

Quantization reduces the numerical precision of model weights from 16-bit or 32-bit floating point to lower bit widths (8-bit, 4-bit, or even 2-bit integers).

Why it works: Most model weights cluster around small values. The difference between representing a weight as 0.0234375 (FP16) versus 0.023 (INT8) is negligible for output quality but halves memory usage.

Common quantization methods:

Method	Bits	Quality Loss	Speed Gain	Memory Reduction
FP16 (baseline)	16	None	1x	1x
INT8 (W8A8)	8	Minimal	1.5-2x	2x
GPTQ (W4A16)	4	Small	2-3x	4x
AWQ	4	Small	2-3x	4x
GGUF Q4_K_M	4	Small	2-3x	4x
QuIP#	2	Moderate	4-5x	8x

Practical example: A 70B parameter model requires ~140GB in FP16, needing 2x A100 80GB GPUs. With 4-bit quantization, it fits on a single A100 or even a consumer RTX 4090 (24GB).

# Quantizing with llama.cpp
./quantize model-f16.gguf model-q4_k_m.gguf Q4_K_M

# Serving with vLLM and AWQ quantization
python -m vllm.entrypoints.openai.api_server \
    --model TheBloke/Llama-3.3-70B-AWQ \
    --quantization awq \
    --tensor-parallel-size 1

Speculative Decoding: Draft and Verify

LLM inference is bottlenecked by sequential token generation — each token requires a full forward pass. Speculative decoding breaks this bottleneck by using a small, fast "draft" model to generate candidate tokens, then verifying them in parallel with the large model.

flowchart TD
    HUB(("Why Inference<br/>Optimization Matters"))
    HUB --> L0["Quantization: Trading<br/>Precision for Speed"]
    style L0 fill:#e0e7ff,stroke:#6366f1,color:#1e293b
    HUB --> L1["Speculative Decoding: Draft<br/>and Verify"]
    style L1 fill:#e0e7ff,stroke:#6366f1,color:#1e293b
    HUB --> L2["KV-Cache Optimization"]
    style L2 fill:#e0e7ff,stroke:#6366f1,color:#1e293b
    HUB --> L3["PagedAttention and vLLM"]
    style L3 fill:#e0e7ff,stroke:#6366f1,color:#1e293b
    HUB --> L4["Continuous Batching"]
    style L4 fill:#e0e7ff,stroke:#6366f1,color:#1e293b
    HUB --> L5["Putting It All Together"]
    style L5 fill:#e0e7ff,stroke:#6366f1,color:#1e293b
    style HUB fill:#4f46e5,stroke:#4338ca,color:#fff

How it works:

1. The draft model (e.g., Llama 3.3 8B) generates K candidate tokens quickly
2. The target model (e.g., Llama 3.3 70B) verifies all K tokens in a single forward pass
3. Accepted tokens are kept; the first rejected token is replaced with the target model's choice
4. The process repeats

Speedup: When the draft model's predictions match the target model (which happens 70-90% of the time for well-chosen pairs), you get K tokens for the cost of ~1 forward pass of the large model. Typical speedups: 2-3x for well-matched model pairs.

KV-Cache Optimization

During autoregressive generation, the Key-Value cache stores computed attention states for all previous tokens. This cache grows linearly with sequence length and can consume more memory than the model weights for long contexts.

Techniques:

• Multi-Query Attention (MQA): Share key/value heads across attention heads, reducing KV-cache by 8-32x
• Grouped-Query Attention (GQA): A middle ground — share KV heads in groups rather than fully
• KV-cache quantization: Compress cached key/value tensors to INT8, halving cache memory
• Sliding window attention: Limit attention to recent tokens plus landmark tokens, capping cache size

PagedAttention and vLLM

PagedAttention, the innovation behind vLLM, manages KV-cache memory the way operating systems manage virtual memory — in non-contiguous pages.

Problem solved: Traditional KV-cache allocation pre-allocates memory based on maximum sequence length, wasting memory for shorter sequences. With batch sizes of 100+ concurrent requests, this waste becomes the primary bottleneck.

How PagedAttention helps:

• Allocates KV-cache in small blocks (pages) on demand
• Eliminates memory waste from pre-allocation
• Enables sharing KV-cache pages across requests using the same prefix (prompt caching)
• Increases throughput by 2-4x compared to naive implementations

# vLLM automatically uses PagedAttention
from vllm import LLM, SamplingParams

llm = LLM(
    model="meta-llama/Llama-3.3-70B-Instruct",
    tensor_parallel_size=2,
    max_model_len=32768,
    gpu_memory_utilization=0.90
)

outputs = llm.generate(
    prompts=["Explain quantum computing" for _ in range(100)],
    sampling_params=SamplingParams(temperature=0.7, max_tokens=512)
)

Continuous Batching

Traditional static batching waits for a full batch before processing and waits for the longest sequence to finish before returning any results. Continuous batching (also called iteration-level batching) inserts new requests and returns completed requests at every generation step.

Impact: Reduces average latency by 50-80% under load and increases throughput by 2-3x compared to static batching. All modern serving frameworks (vLLM, TGI, TensorRT-LLM) implement continuous batching by default.

Putting It All Together

A production-optimized inference stack combines multiple techniques:

Request → Continuous Batching Engine
            ├── PagedAttention (memory efficiency)
            ├── Quantized Model (INT8/INT4)
            ├── GQA/MQA (reduced KV-cache)
            ├── Speculative Decoding (speed)
            └── Prefix Caching (shared prompts)

The compound effect of these optimizations is dramatic: a well-optimized serving stack can serve 10-50x more requests per GPU compared to a naive implementation, reducing per-query costs proportionally.

---

Sources: vLLM — PagedAttention Paper, Hugging Face — Quantization Guide, DeepSpeed — Inference Optimization

flowchart LR
    IN(["Input prompt"])
    subgraph PRE["Pre processing"]
        TOK["Tokenize"]
        EMB["Embed"]
    end
    subgraph CORE["Model Core"]
        ATTN["Self attention layers"]
        MLP["Feed forward layers"]
    end
    subgraph POST["Post processing"]
        SAMP["Sampling"]
        DETOK["Detokenize"]
    end
    OUT(["Generated text"])
    IN --> TOK --> EMB --> ATTN --> MLP --> SAMP --> DETOK --> OUT
    style IN fill:#f1f5f9,stroke:#64748b,color:#0f172a
    style CORE fill:#ede9fe,stroke:#7c3aed,color:#1e1b4b
    style OUT fill:#059669,stroke:#047857,color:#fff

flowchart TD
    HUB(("Why Inference<br/>Optimization Matters"))
    HUB --> L0["Quantization: Trading<br/>Precision for Speed"]
    style L0 fill:#e0e7ff,stroke:#6366f1,color:#1e293b
    HUB --> L1["Speculative Decoding: Draft<br/>and Verify"]
    style L1 fill:#e0e7ff,stroke:#6366f1,color:#1e293b
    HUB --> L2["KV-Cache Optimization"]
    style L2 fill:#e0e7ff,stroke:#6366f1,color:#1e293b
    HUB --> L3["PagedAttention and vLLM"]
    style L3 fill:#e0e7ff,stroke:#6366f1,color:#1e293b
    HUB --> L4["Continuous Batching"]
    style L4 fill:#e0e7ff,stroke:#6366f1,color:#1e293b
    HUB --> L5["Putting It All Together"]
    style L5 fill:#e0e7ff,stroke:#6366f1,color:#1e293b
    style HUB fill:#4f46e5,stroke:#4338ca,color:#fff