魁匠团这个做吃鸡的公司居然也开源模型了，29B 参数 MoE，韩英双语能力不错，感兴趣可以去试试。
韩国游戏公司魁匠团（Krafton）开源了 MoE 模型 Raon-1-mini-preview，总参数量 29B，激活参数 4.3B。据发布信息，其整体评分高于 Qwen3.6 35BA3B。该模型对韩语和英语的支持比较突出。模型已在开源平台放出。
Raon-1-mini-preview：韩国游戏公司魁匠团开源的MoE模型
29B总参数，4.3B激活参数，整体评分比Qwen3.6 35BA3B高，对韩语和英语的支持比较好。
模型：https://t.co/Buf8IVhioP https://t.co/O3ya97x0oN
Day-0 support means the model can be self-hosted on existing vLLM infrastructure immediately, without waiting for community integrations. With only 3B active parameters, it targets teams running agentic workloads on modest local hardware who want an alternative to larger dense models. Developers comparing open-source MoE options for on-premise serving should evaluate its throughput against Qwen and DeepSeek MoE checkpoints.
米哈游和清华北大把游戏操作模型开源了，能自己玩《我的世界》和《毁灭战士》，最小 0.8B 单卡就能跑，可以拿来折腾。
Training agents on verifiable, evidence-grounded tasks addresses the core bottleneck of RL for long-horizon work: reward reliability. The 16.7K-environment scale makes this one of the larger open efforts in agent environment construction. Engineers building work or office agents can study the dual-verifier design for their own training pipelines.
中电信星辰实验室开源了 29B 的意图理解专用模型 Xing4.0-29B-Intent，做对话系统和意图识别的可以拿来试试。
Distilling from a large teacher into a small model specialised for long tool-use chains may matter more than raw size for local agents, a trade-off the OSWorld gap makes visible. Builders running agents on local hardware can use it as a drop-in for multi-tool MCP workflows without cloud dependency. Compared with Qwen3.6-27B, it wins on tool orchestration but loses on general OS-level tasks.
OpenAI 用 GPT-5.6 Sol 挖出了影响全球 70% 网站的 PHP 核心漏洞，还和另一团队协同修了两个月，AI 安全研究的新玩法。
