# Moonshot AI releases Kimi K3.1 Pro with open weights due by Nov 4, 2026

> Original title: 月之暗面开源级 3T 模型升级：Kimi K3.1 Pro 强化编程、3D 与金融 Agent

- Source: meng shao
- Published: 2026-10-24T01:05:37.000Z
- Bad Influence: https://badinfluence.news/items/tw0xz7h2c5t6dzbl6f5n5bku4
- Original: https://x.com/shao__meng/status/2113799036004450807

## Summary

Moonshot AI launched Kimi K3.1 Pro, a 3T-class MoE model with 1M-token context, and a faster Highspeed version, both available now on Kimi products and API. The quoted post says Terminal-Bench 4.0 scores 65.7%, near Claude Opus 5.5's 66.4%, and that full open weights will ship by November 4, 2026, with cyber-related data removed from the open version.

## Post

月之暗面开源级 3T 模型升级：Kimi K3.1 Pro 强化编程、3D 与金融 Agent

K3.1 Pro 是 K3 的大规模后训练版，号称目前最强的 3T 级开源权重模型（1M token 上下文）；在 Terminal-Bench 4.0 等 Agent 编程榜上接近 Claude Opus 5.5，并同步放出 2–3× 速度的 Highspeed 版，完整权重计划 2026-11-04 前开源。
https://www.kimi.ai/blog/kimi-k3-1-pro

模型定位
· Kimi K3.1 Pro：K3.1 系列首发，基于 Kimi K3 加大后训练算力，强调跨数百下游域的联合 RL。
· Kimi K3.1 Pro Highspeed：同能力线，交互速度约 2–3×。
· 今日起可用：http://Kimi.ai、Kimi Work、Kimi Code、Kimi API；完整权重 11 月 4 日前发布。企业端默认关闭，需管理员开启。

核心技术叙事
团队优化了整条 RL 栈（框架、长上下文推理、训练效率、沙箱、rollout 调度、环境质检平台），使每单位 RL 算力回报更稳定，从而在各域「一致提升」。

编程与长程 Agent
· Terminal-Bench 4.0：65.7%（用 Claude-Code harness），高于 GPT-6 Astra 58.2%，略低于 Claude Opus 5.5 的 66.4%。
· Kimi Code Bench 3.0 Long-horizon（内部）：61.2%，高于 K3 的 55.95%。
· 覆盖修 bug 到整项目构建；Unity / Unreal / Godot + 搜索 + 3D 生成；前端与 Three.js 观感更好。
· 内部案例：基于含 296 骨骼件、402 肌肉的 human-atlas，做人体行走动力学模拟；167 次任务、420 次独立评审后，连续行走约 11.2 秒；奖励塑形无效时，用开环探针查出并修正肩部肌力矩臂 36 处错误后再训练。

3D 与设计
· BenchCAD（多视图渲染 → CAD 代码）：带工具 92.7%，接近 Opus 5.5（96.2%）与 GPT-6 Astra（95.9%）。
· 强项：参数化建模（机械件、建筑、可配置产品、定制家具）；无需 MCP 也可写/跑 Blender、CAD 脚本。
· 产品侧：3D Make 已上线 http://kimi.ai 与 App，可在线预览编辑、导出打印或视频/Live Photo。

金融
· OfficeQA Pro v2：64.4，文称 SOTA（复杂财报检索、跨文档校验、多步数值推理）。
· 相对 K3：Financial Modeling +7.2；Financial Research +11.6；Business Operation 约 4.2×。
· 幻觉：Kimi Hallu Bench Hard 上 claim 级幻觉率 23.4% → 8.3%；含至少一处事实错误的回复比例约为前代的 42.4%（压力测试，不代表日常整体错误率）。

网络安全与开源原则（篇幅最长的一节）
· 观察到：后训练放大后，漏洞研究能力随算力上升；溯源到 mid-training 语料与 RL 环境里的 cyber 相关内容。
· 开源权重版主动从 mid-training 与 RL 两侧剔除 cyber 相关数据，保留通用编程任务。
· xdev-bench-hard（内核、Node、JSC、Exim、VirGL 等近期漏洞复现，需写利用拿 flag）：未做风险削减 checkpoint pass@3 = 90.5%；最终 K3.1 Pro = 81.0%。
· 立场：不把「进攻性 cyber 绝对前沿」放进开源权重；同时认为逐步扩散双用途能力（尤其防御与自审代码）长期更安全。能力仍落后于已公开至少三个月的头部闭源模型。

## AI translation

Moonshot AI open-sources a 3T-class model upgrade: Kimi K3.1 Pro strengthens programming, 3D, and finance Agents

K3.1 Pro is a large-scale post-training version of K3, billed as the strongest open-weight model at the 3T level (1M token context). On Agent programming leaderboards such as Terminal-Bench 4.0 it comes close to Claude Opus 5.5, and it also releases a Highspeed version at 2–3× speed. The full weights are planned to be open-sourced before 2026-11-04.
https://www.kimi.ai/blog/kimi-k3-1-pro

Model positioning
· Kimi K3.1 Pro: the debut of the K3.1 series, built on Kimi K3 with more post-training compute, emphasizing joint RL across hundreds of downstream domains.
· Kimi K3.1 Pro Highspeed: same capability line, with interaction speed about 2–3×.
· Available starting today: http://Kimi.ai, Kimi Work, Kimi Code, Kimi API; full weights released before November 4. Off by default on the enterprise side, requires an admin to enable.

Core technical narrative
The team optimized the entire RL stack (framework, long-context inference, training efficiency, sandboxes, rollout scheduling, environment quality-control platform), making the return per unit of RL compute more stable, thereby achieving "consistent improvement" across domains.

Programming and long-horizon Agents
· Terminal-Bench 4.0: 65.7% (using the Claude-Code harness), higher than GPT-6 Astra's 58.2%, slightly lower than Claude Opus 5.5's 66.4%.
· Kimi Code Bench 3.0 Long-horizon (internal): 61.2%, higher than K3's 55.95%.
· Covers everything from fixing bugs to building whole projects; Unity / Unreal / Godot + search + 3D generation; better front-end and Three.js visual quality.
· Internal case: using a human-atlas containing 296 bone pieces and 402 muscles, it simulated human walking dynamics; after 167 tasks and 420 independent reviews, it achieved continuous walking for about 11.2 seconds; when reward shaping was ineffective, it used open-loop probes to find and correct 36 errors in shoulder muscle moment arms before retraining.

3D and design
· BenchCAD (multi-view rendering → CAD code): 92.7% with tools, close to Opus 5.5 (96.2%) and GPT-6 Astra (95.9%).
· Strengths: parametric modeling (mechanical parts, architecture, configurable products, custom furniture); can write and run Blender and CAD scripts without MCP.
· Product side: 3D Make is live on http://kimi.ai and the App, allowing online preview and editing, and export to print or video/Live Photo.

Finance
· OfficeQA Pro v2: 64.4, claimed as SOTA (complex financial report retrieval, cross-document verification, multi-step numerical reasoning).
· Compared with K3: Financial Modeling +7.2; Financial Research +11.6; Business Operation about 4.2×.
· Hallucination: on Kimi Hallu Bench Hard, claim-level hallucination rate dropped from 23.4% to 8.3%; the proportion of responses containing at least one factual error is about 42.4% of the previous generation's (stress test, not representative of overall everyday error rates).

Cybersecurity and open-source principles (the longest section)
· Observation: after post-training scaling, vulnerability research capability rises with compute; traced to cyber-related content in the mid-training corpus and RL environments.
· The open-weight version proactively removes cyber-related data from both mid-training and RL, while retaining general programming tasks.
· xdev-bench-hard (reproducing recent vulnerabilities in the kernel, Node, JSC, Exim, VirGL, etc., requiring writing exploits to capture flags): without risk-mitigation checkpoint pass@3 = 90.5%; final K3.1 Pro = 81.0%.
· Stance: does not put "absolute frontier offensive cyber capability" into open weights; at the same time, believes that gradually spreading dual-use capabilities (especially defense and self-auditing of code) is safer in the long run. Its capability still lags behind top closed-source models that have been public for at least three months.

![](https://badinfluence.news/api/img-proxy?u=https%3A%2F%2Fpbs.twimg.com%2Fmedia%2FHVW4-s0aMAAxPgI.jpg&mode=thumb&exp=1793404800&sig=e4ed811e8c9600fd)

## Quoted post @Kimi_Moonshot

> Introducing Kimi K3.1 Pro: our most capable 3T-class open-weight model.
> 
> 🔹 1M-token context, 2.8T total / 104B active params (MoE), Multimodal.
> 🔹 Full-project coding: games in Unreal, Unity & Godot.
> 🔹 Excels at parametric 3D modeling in Blender & CAD.
> 🔹 More reliable at finance work: Accurate modeling with hallucination down by 65%.
> 🔹 Matching the strongest proprietary models across coding, agentic and knowledge benchmarks.
> 
> Kimi K3.1 Pro and K3.1 Pro Highspeed (2–3× faster) are now live on http://Kimi.ai, Kimi Work, Kimi Code, and the Kimi API.  
> 
> Open weights by Nov 4, 2026.
> 
> 🔗 API: http://platform.kimi.ai 
> 🔗 Tech blog: http://kimi.ai/blog/kimi-k3-1-pro

https://x.com/Kimi_Moonshot/status/2113680701770068067