Skip to content

Models layerModel Economics

Model distillation

What it is

Training a small, cheap model to copy a big, expensive one so it can do similar work at a fraction of the cost.

3companies exposed1part it leans on2parts need it
Where Model distillation sitsLayer 4 of 5
EnergyChipsInfraModelsAppsModel distillation
Model relevance51/100

Ranks moderate (51/100) - set apart by rising momentum (72) and cross-layer reach (66).

The short read

Why Model distillation matters, and what breaks without it.

Why it matters

It is the primary lever converting expensive frontier models into economically servable products at scale.

Why now

2025's low-cost reasoning models showed that distilling large-model behavior into small ones can approach frontier quality at a fraction of the serving cost, forcing a repricing of the field.

If Model distillation runs short

Over-distillation loses edge-case capability and safety behavior, shipping cheap models that fail silently on hard inputs.

Editorial read

In depth · editorial + model

Compressing a large teacher model's behavior into a smaller student to slash serving cost and latency with limited quality loss. It is the primary lever converting expensive frontier models into economically servable products at scale. 2025's low-cost reasoning models showed that distilling large-model behavior into small ones can approach frontier quality at a fraction of the serving cost, forcing a repricing of the field. Over-distillation loses edge-case capability and safety behavior, shipping cheap models that fail silently on hard inputs.

Mental models

How to think about it

  1. 01Capability is created once, served many ways
  2. 02Cheap intelligence is distilled, not always trained fresh

Signals to follow

What to watch

  • Small-model quality-vs-cost frontier
  • Terms-of-service / IP disputes over distilling competitor outputs
  • Distilled reasoning-model releases

Questions

Frequently asked

The short answers about Model distillation, from the same model as the rest of the page.

What is Model distillation?

Training a small, cheap model to copy a big, expensive one so it can do similar work at a fraction of the cost.

Why does Model distillation matter for AI?

It is the primary lever converting expensive frontier models into economically servable products at scale.

Who makes Model distillation?

The companies the model tags as producers or suppliers of Model distillation: Alphabet (Google), OpenAI, Anthropic.

Which companies are most exposed to Model distillation?

Alphabet (Google), OpenAI, Anthropic. In total, 3 companies are mapped to Model distillation.

What happens if Model distillation runs short?

Over-distillation loses edge-case capability and safety behavior, shipping cheap models that fail silently on hard inputs.

Where does Model distillation sit in the AI value chain?

Model distillation sits in the Models layer of the AI value chain.

In the app

Go deeper on Model distillation

Inside THE ENTITY, Model distillation is live: its pressure, its links and the companies on it, refreshed nightly.

  • The materials, geographies and policies it depends on, heat-mapped
  • Substitutes, relief valves and the domino chains if it tightens
  • The live tension score, momentum and news drivers
  • Four levels of analysis, from plain English to strategic
The whole chain

Research, not investment advice. Model scores are illustrative reads from our model of the AI value chain, and companies are mapped to explain the chain, never as recommendations.

as of 2026-11-09Medium confidence model v2.01