Revision History
Kimi K2 · 10 revisions
Sizes are character counts of the article source. The signed number is the change from the previous revision.
Recent edit summaries
Detailed summaries recorded by editors. Generic maintenance summaries are omitted here; every recorded revision remains below. Dates describe the edit, not necessarily the event it covers.
- Correction: Corrected fabricated quotes attributed to the Kimi K2 technical report and Nathan Lambert, Moonshot's launch API price, an unsupported $2 billion/$20 billion funding claim, the Kimi Linear release date, misstated technical-report rationales for expert count and attention heads, competitor benchmark and parameter figures, a claim that DeepSeek-V3 used Muon, a misreading of NIST CAISI findings, the claim that K2.6 added video input, Composer 1 being built on K2, and stale statements that Kimi K3 had not been released (plus 33 minor fixes)
Version 10 · Nov 2, 2026, 12:40 PM
- Update: added the developer's official video.
Version 9 · Nov 1, 2026, 06:40 AM
- Correction: GLM-4.6 uses grouped-query attention (96 query heads, 8 KV heads) with 160 routed experts and 8 active, not MLA.
Version 8 · Oct 31, 2026, 03:47 AM
- Improve internal links (2026-10-13)
Version 6 · Oct 13, 2026, 10:20 AM
- Added 6 contextual internal links
Version 5 · Jul 23, 2026, 03:29 PM