Skip to content

Generative AI & LLMs

Speculative decoding

Speeding up AI writing by having one system propose pieces that the main AI checks before accepting.

Example

A smaller AI drafts a continuation that the main AI checks.

Why people use it

It can speed up writing by letting a faster system propose pieces for a larger one to check.

What you'll hear

“Let the smaller system draft the next few pieces.”

What this means for you

Measure the whole response time with the actual service and task.

Can you control it?

Developer-only

The people building or running the AI choose this setup. An everyday user generally needs their help to change how this part works.

Common questions

Does the draft AI system alone determine the final result?
No. The target checking procedure controls which draft tokens are accepted.
Does a rejected draft mean the request fails?
No. The larger system can continue with different pieces instead.
Will it speed up every request equally?
No. The benefit depends on how many proposed pieces are accepted and the computing setup.

Related terms

Still have questions?

Up to 500 characters.

Ask LATHIC about AI. Relevant glossary entries may be included.

Your question, the glossary entries it matches, and a rotating pseudonymous identifier go to Microsoft Azure’s OpenAI service through Vercel AI Gateway to generate an answer. Zero retention and no training are required of the provider, and LATHIC does not save your question or answer. Privacy Notice