Open Models Almanac
Methodsinference & servingcontext management

specific method · filed under inference & serving

Context compaction

Compacts context at a stated 300K-token threshold; the evidence does not specify the compaction operation.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See context management for the methods filed beside it.

Also called context compaction at 300K tokens, Context-compaction strategy.

sources
3
model
1
lab adopt it
1
strongest
used

Documented in

Moonshot AI

Evidence

3 quotes from 3 sources, one card per model, each opening with its strongest.

How sources treat it
used 3

Kimi K3

Moonshot AIadopted3 sources

We adopt the context-compaction strategy used in the Claude model cards, triggered at 300K tokens.

usedevaluation onlyin Kimi K3Moonshot AI
2 more from this model

We adopt a context-compaction strategy triggered at 300K tokens.

usedinference servingin Kimi K3Moonshot AI

We adopt a context-compaction strategy triggered at 300K tokens.

usedevaluation onlyin Kimi K3Moonshot AI

Further reading

Picked by hand, not extracted: where to read more, not evidence for anything on this page.

↑