# O12: Speculative Decoding, v0.21.0

- Publisher: vLLM
- URL: [Primary source](https://docs.vllm.ai/en/v0.21.0/features/speculative_decoding/)
- Published: not stated
- Updated/version date: not established
- Accessed: 2026-10-22
- Type: versioned official runtime documentation
- Verification: opened source, rather than relying on a search excerpt

## Supported finding

The documented target includes reducing inter-token latency for medium-to-low-QPS, memory-bound workloads.

## Limits and transfer

Version-pinned mechanism example, not a claim about the latest release or universal throughput improvement.

## Use in the session

Mechanism and decision conditions. Retain the qualifications when translating the finding into slide copy. No source establishes Knauf's savings.