Kimi K3.1 Pro model weights are out

📰 Tech Blog | 🚀 API Quickstart

📰 技术博客 | 🚀 API 快速入门

Kimi K3.1 Pro is an open-weight, natively multimodal agentic model, and our most capable 3T-class model to date. Post-trained from Kimi K3 with large-scale joint reinforcement learning across hundreds of downstream domains, it is the first model in the K3.1 series. With a 1-million-token context window, Kimi K3.1 Pro delivers substantial improvements in coding, 3D creation, financial research, factual accuracy, and other agentic capabilities.

Kimi K3.1 Pro 是一款开放权重的原生多模态智能体模型，也是我们迄今为止能力最强的 3T 级别模型。它基于 Kimi K3 进行后训练，通过跨越数百个下游领域的大规模联合强化学习而成，是 K3.1 系列的首款模型。凭借 100 万 token 的上下文窗口，Kimi K3.1 Pro 在编程、3D 创作、金融研究、事实准确性以及其他智能体能力方面均实现了显著提升。

Compared with Kimi K3, we replace the full-attention Gated MLA layers with Gated DeepSeek Sparse Attention (Gated DSA) layers for more efficient long-context modeling.
与 Kimi K3 相比，我们将全注意力 Gated MLA 层替换为 Gated DeepSeek Sparse Attention (Gated DSA) 层，以实现更高效的长上下文建模。

Unless otherwise specified, all Kimi K3.1 Pro results use reasoning effort max, temperature = 1.0, and top-p = 0.95. A dash (—) means that the source table does not report a score. Higher scores are better except where the benchmark name explicitly states lower is better; Elo scores and other benchmark-specific metrics are not directly comparable with accuracy or pass-rate scores.

除非另有说明，所有 Kimi K3.1 Pro 的结果均使用 reasoning effort max、temperature = 1.0 和 top-p = 0.95。破折号 (—) 表示源表格未报告分数。除基准测试名称明确指出分数越低越好外，分数越高越好；Elo 分数和其他特定于基准测试的指标不能直接与准确率或通过率分数进行比较。

You can access Kimi K3.1 Pro’s API on https://platform.kimi.ai by selecting kimi-k3.1-pro, and we provide OpenAI/Anthropic-compatible API for you. Currently, Kimi K3.1 Pro is recommended to run on the following inference engines:

您可以通过在 https://platform.kimi.ai 选择 kimi-k3.1-pro 来访问 Kimi K3.1 Pro 的 API，我们为您提供兼容 OpenAI/Anthropic 的 API。目前，建议在以下推理引擎上运行 Kimi K3.1 Pro：

The thinking effort of Kimi K3.1 Pro is configured with the top-level reasoning_effort request field, which supports “low”, “high”, and “max” (default “max”).

Kimi K3.1 Pro 的思考努力程度通过顶层的 reasoning_effort 请求字段进行配置，支持 “low”、“high” 和 “max”（默认为 “max”）。

For multi-turn conversations and tool calls, Kimi K3.1 Pro requires the complete assistant message returned by the API to be passed back to messages as-is — including reasoning_content and tool_calls, not just content:

对于多轮对话和工具调用，Kimi K3.1 Pro 要求将 API 返回的完整助手消息原样传回 messages——包括 reasoning_content 和 tool_calls，而不仅仅是 content：

For full guides and examples (vision input, structured output, partial mode, tool choice, dynamic tool loading, context caching), see the Kimi K3.1 Pro Quickstart and Thinking Effort.

有关完整指南和示例（视觉输入、结构化输出、部分模式、工具选择、动态工具加载、上下文缓存），请参阅 Kimi K3.1 Pro 快速入门和思考努力程度说明。

Kimi K3.1 Pro is released under the Kimi K3.1 Pro License.

Kimi K3.1 Pro 在 Kimi K3.1 Pro 许可协议下发布。

If you have any questions, please reach out at support@moonshot.ai.

如有任何疑问，请通过 support@moonshot.ai 与我们联系。