什么是 Kimi?
Kimi 以 K 系列通用旗舰持续迭代,保留 K2、K2 Thinking、K2.5、K2.6 至 K3 的历史;Kimi Code 使用独立代码入口和型号。
适合软件工程、资料整理和长程知识工作;Kimi App、Code、API 与开放权重的限制分别核对。
适合软件工程、资料整理和长程知识工作;Kimi App、Code、API 与开放权重的限制分别核对。
Kimi 以 K 系列通用旗舰持续迭代,保留 K2、K2 Thinking、K2.5、K2.6 至 K3 的历史;Kimi Code 使用独立代码入口和型号。
先看各产品线做什么、当前有哪些代表版本,再理解整个家族的能力与边界。
Kimi 以 K 系列通用旗舰持续迭代,保留 K2、K2 Thinking、K2.5、K2.6 至 K3 的历史;Kimi Code 使用独立代码入口和型号。
适合软件工程、资料整理和长程知识工作;Kimi App、Code、API 与开放权重的限制分别核对。
适合软件工程、资料整理和长程知识工作;Kimi App、Code、API 与开放权重的限制分别核对。
先从代表性任务理解用途;更完整的选择条件请切换产品模式。
选择具体版本后,用真实业务样本核对正确率、响应时间和单位任务成本。
选择具体版本后,用真实业务样本核对正确率、响应时间和单位任务成本。
选择具体版本后,用真实业务样本核对正确率、响应时间和单位任务成本。
用事实核对模型能力、价格和接入条件;更多记录可按关键词、主题和分页查阅。
K2 旧预览端点停止服务,不再维护和支持。 适用范围:官方 API 预览端点;不改写开源权重可用性。
2026-05-25: Legacy K2 preview endpoints retire. Scope: Official API preview endpoints, not weight availability.
调整为 $0.15/$1.15/$8;11 月 6 日生效,7 日发布说明。 适用范围:当时的 kimi-k2-turbo API;历史事件,非当前在售价格。
2025-11-06: Rates become $0.15/$1.15/$8; effective November 6, notice November 7. Scope: kimi-k2-turbo API at that time; not current catalog rates.
官方公告再次提升 Turbo 预览输出速度;不补写未核实倍数。 适用范围:当时的 kimi-k2-turbo-preview 官方 API。
2025-08-22: Official notice reports another Turbo-preview speed improvement; no unverified multiplier. Scope: Official kimi-k2-turbo-preview API at that time.
同一家族里的产品线各有定位;从这里进入具体版本。
通用推理、多模态与长时程 Agent 的连续迭代
软件开发与高速交互
所有 AI 回答均基于已审核数据与来源证据,事实 / 推断 / 未核验分开呈现。
这些是本页引用的资料;价格、渠道和规格请进入具体版本核对,并以官方页面为准。
2026-09-08: Ark Coding Plan supports K3 with a high credit factor; recommends Pro subscribers. Scope: Ark Coding Plan subscribers, not first-party API launch.
2026-08-31: Official API support ends; open weights remain a separate channel. Scope: K2.5 users on the official Kimi API.
2026-02-11: Ark Coding Plan adds K2.5 visual understanding. Scope: K2.5 on Ark, not its first model release.
2025-11-06: Rates become $0.15/$1.15/$8; effective November 6, notice November 7. Scope: kimi-k2-turbo API at that time; not current catalog rates.
K2 的 0905 更新,单独保留初版与 9 月版本。
2025-08-22: Official notice reports another Turbo-preview speed improvement; no unverified multiplier. Scope: Official kimi-k2-turbo-preview API at that time.
Kimi K2.7 Code June 12, 2026 Kimi K2.7 Code, our latest coding model, is now released and open-sourced — and available in Kimi For Coding. The new model takes effect only with Thinking On (thinking mode enabled). Improved coding & agent performance over K2.6 +10.4% on Program-Bench, +11.4% on MCP Mark Verified, and +76.2% on SWE Marathon. Reasoning efficiency Less overthinking, with 30% lower reasoning-token usage compared to K2.6. Long-horizon coding Improved instruction following and higher end-to-end coding task success rates. ⚡ Kimi K2.7 Code HighSpeed Edition (beta coming soon) The same model at roughly 5-6× the output speed. Kimi Code CLI v0.14.0 June 10, 2026 Interrupt hook event When you interrupt a turn (for example by pressing Esc), hooks now receive a dedicated Interrupt event — it fires in place of Stop, so external tooling no longer mistakes an interrupted turn for one that is still running. Programmatic interruptions such as timeouts don't trigger it. → Event reference Kimi Datasource adds legal data The official data plugin is now at v3.2.0, adding Chinese laws, regulations, and judicial cases — coverage now spans five data domains: stock market, macr