Repository Issue Activity (beta)

#41 Release Kimi K3 under a FLOSS license

#40 UX improvement: Provide direct and equally visible “Continue with Google” and “Continue with phone number” options

#38 [Critical] Kimi Work 3.0 Orchestrator burns ~$150 in 15 minutes on trivial task – ignores explicit constraints (Incident: Aug 18, 2026)

#37 [Bug/Service] Coding Plan quota exhausted due to model infinite loop. Rejecting invalid "Premium Membership" compensation and requesting full refund.Coding Plan 额度因模型死循环异常耗尽，拒绝无效的“高级会员”补偿，要求全额退款

#14 Why would people want to use this 😅

#36 kimi major and minor bug audits

#35 HOW TO IMPROVE TOKEN EFFICIENCY BY KIMI

#34 [Discussion] A Few Thoughts on the Path to AGI

#33 How can I run the Kimi K3 model on multiple AMD MI308X servers using the vLLM or SGLang framework?

#32 Disbalance in the usage budget

#31 [Feature Request] Transparent "Auto Max" mode — K3 for planning/verification, K2.6 for execution

#30 Need license for Deutsche Telekom AG

#28 探讨一下最核心的地方

#6 这么一个强大的模型，真希望README.md是用中文来呈现的

#27 kimi-k3 token usage counts completely messed up on openrouter / third-party routers smh

#26 KV-Cache Memory Spike and OOM during High-Context Prefilll (1M Token Window) in Kimi Delta Attention Layer

#22 Verify evals on Papers with Code

#19 kimi真的太强了！！！！！

#18 在一台mac mini 4 上如何跑起来

#12 K3 tool call error repeat and repeat

#13 Release Kimi K3 on Hugging Face

#5 [Request] Apple Silicon / MLX / EXO support roadmap

#10 基于中间层隐状态读取的内生安全通路——替代外挂护栏与参数对齐的第三种架构

#7 是否考虑出个27b或70B的？

#8 技术报告无法获取了？