{"data":{"slug":"kimi-k2","etype":"model","name":"Kimi K2","cat":"Language","subtitle":"A trillion-parameter open MoE tuned for agentic work.","summary":"Moonshot's K2 (July 2025) is a 1T-parameter MoE (32B active) released with open weights and explicitly optimized for tool use and agentic coding; K2 Thinking (November 2025) added a reasoning tier that challenged closed frontier models on several benchmarks. Trained under export-control constraints, it is among the largest open models ever released. K2.6 and a K3 generation followed through 2026 (per ecosystem roundups).","attrs":null,"kf":{"params":"1T (32B active)","context":"256K","weights":"Open","modality":"Text","rel_date":"2025-07","developer":"Moonshot AI"},"aliases":["Kimi","K2","K2.6","K3"],"ticker":null,"exchange":null,"hq":null,"founded":null,"website":null,"status":"Current","region":null,"asof":"2026-01-15","sources":[{"url":"https://moonshotai.github.io/Kimi-K2/","label":"Kimi K2"}]},"meta":{"found":true,"count":1,"served_from":"site cache (no live upstream calls)","as_of":"2026-10-14T08:41:58+00:00","disclaimer":"Research and information only — not investment advice. Structured records carry basis labels (vendor spec / reported / estimate); market quotes are delayed/indicative.","attribution":"Records compiled from vendor documentation, official filings and reputable public reporting; market quotes via financial data providers. See /data-sources."}}