Kimi Teases K3.1 as Anonymous Arena Model Kibitz Draws Coding Tests
Kimi posted a 24-second teaser showing Kimi Code with “Model: Kimi K3.1” displayed, but did not announce a release date, parameter count, or pricing. An anonymous Arena model named Kibitz has produced frontend and 3D coding demonstrations. Community observers inferred that Kibitz may be Kimi K3.1, partly based on tokenizer characteristics, but this identity remains unconfirmed. Informal comparisons were mixed: Kibitz completed one voxel-courtyard test faster than GPT-6 Astra Max, while another tester favored Gemini 4 Carbon over Kibitz, Kalmia, and Claude Opus 5.5.
The official teaser indicates that coding and developer workflows are central to the K3.1 update. The Arena results provide only preliminary observations about Kibitz; if Kibitz is confirmed as K3.1, they may serve as early indicators of its coding performance.
本月底之前，国内大模型阵营里最让人惦记的两款旗舰——Kimi K3.1与智谱 GLM-5.5——正陆续浮出水面，其中 Kimi K3.1已经悄悄在 Arena 竞技场上跑起了测试，代号 Kibitz。
多个网友顺着分词器等蛛丝马迹，基本确认了 Kibitz 就是 Kimi K3.1。此前的爆料勾画的升级方向很清晰:主攻长程自主任务，强化游戏与3D 代码能力，并补齐操作电脑的本事，一整套打法明显是对标 OpenAI 等前沿模型的"一线玩法"。
今天多位网友实打实上手试了一圈，反馈相当炸裂——无论是鹈鹕这类刁钻图像，还是3D 动画的生成，K3.1都显示出能与 GPT-6Astra 掰手腕的实力。按 Arena 的惯例，在擂台上测试的大模型通常会在一周后正式发布，所以本月底依然有希望等到 Kimi K3.1的官宣上线。
市场给它的预期不低:K3.1大概率会成为国产 AI 大模型的新标杆，也有能力挤进全球 TOP3;在 AA 榜单上，它被寄望成为首个突破50分的国产模型，理想状态能摸到55分甚至60分上下。不过性能之外，K3.1还得补上可用性与性价比这两门课——上一代 K3就因为算力资源吃紧，落得速度偏慢、套餐给的额度也不高，过去两个月里声量被 DeepSeek V4.1Flash 这类又快又便宜的模型压过一头。当国产旗舰准备正面对标 Astra，光把分数跑漂亮还不够，能不能让开发者用得顺、用得起，才是它真正翻盘的关键。
#Kimi K3.1#Kimi#Moonshot AI#Kimi Code#Kibitz#GPT-6 Astra Max#Gemini 4 Carbon#Claude Opus 5.5#SGLang