[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-75132":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":9,"language":9,"languages":9,"totalLinesOfCode":9,"stars":10,"forks":11,"watchers":12,"openIssues":13,"contributorsCount":14,"subscribersCount":14,"size":14,"stars1d":15,"stars7d":16,"stars30d":17,"stars90d":14,"forks30d":14,"starsTrendScore":18,"compositeScore":19,"rankGlobal":9,"rankLanguage":9,"license":20,"archived":21,"fork":21,"defaultBranch":22,"hasWiki":21,"hasPages":21,"topics":23,"createdAt":9,"pushedAt":9,"updatedAt":24,"readmeContent":25,"aiSummary":26,"trendingCount":14,"starSnapshotCount":14,"syncStatus":27,"lastSyncTime":28,"discoverSource":29},75132,"Kimi-K2.5","MoonshotAI\u002FKimi-K2.5","MoonshotAI","Moonshot's most powerful model",null,2024,257,19,42,0,11,20,76,33,96.83,"Other",false,"master",[],"2026-06-12 04:01:17","\u003Cdiv align=\"center\">\n  \u003Cpicture>\n      \u003Cimg src=\"figures\u002Fkimi-logo.png\" width=\"30%\" alt=\"Kimi K2.5\">\n  \u003C\u002Fpicture>\n\u003C\u002Fdiv>\n\u003Chr>\n\u003Cdiv align=\"center\" style=\"line-height:1\">\n  \u003Ca href=\"https:\u002F\u002Fwww.kimi.com\" target=\"_blank\">\u003Cimg alt=\"Chat\" src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F🤖%20Chat-Kimi%20K2.5-ff6b6b?color=1783ff&logoColor=white\"\u002F>\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Fwww.moonshot.ai\" target=\"_blank\">\u003Cimg alt=\"Homepage\" src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FHomepage-Moonshot%20AI-white?logo=Kimi&logoColor=white\"\u002F>\u003C\u002Fa>\n\u003C\u002Fdiv>\n\n\u003Cdiv align=\"center\" style=\"line-height: 1;\">\n  \u003Ca href=\"https:\u002F\u002Fhuggingface.co\u002Fmoonshotai\" target=\"_blank\">\u003Cimg alt=\"Hugging Face\" src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002F%F0%9F%A4%97%20Hugging%20Face-Moonshot%20AI-ffc107?color=ffc107&logoColor=white\"\u002F>\u003C\u002Fa>\n  \u003Ca href=\"https:\u002F\u002Ftwitter.com\u002Fkimi_moonshot\" target=\"_blank\">\u003Cimg alt=\"Twitter Follow\" src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FTwitter-Kimi.ai-white?logo=x&logoColor=white\"\u002F>\u003C\u002Fa>\n    \u003Ca href=\"https:\u002F\u002Fdiscord.gg\u002FTYU2fdJykW\" target=\"_blank\">\u003Cimg alt=\"Discord\" src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FDiscord-Kimi.ai-white?logo=discord&logoColor=white\"\u002F>\u003C\u002Fa>\n\u003C\u002Fdiv>\n\u003Cdiv align=\"center\" style=\"line-height: 1;\">\n  \u003Ca href=\"LICENSE\">\u003Cimg alt=\"License\" src=\"https:\u002F\u002Fimg.shields.io\u002Fbadge\u002FLicense-Modified_MIT-f5de53?&color=f5de53\"\u002F>\u003C\u002Fa>\n\u003C\u002Fdiv>\n\u003Cp align=\"center\">\n\u003Cb>📰  \u003Ca href=\"https:\u002F\u002Fwww.kimi.com\u002Fblog\u002Fkimi-k2-5.html\">Tech Blog\u003C\u002Fa>\u003C\u002Fb> |     \u003Cb>📄  \u003Ca href=\"tech_report.pdf\">Full Report\u003C\u002Fa>\u003C\u002Fb>\n\u003C\u002Fp>\n\n## 1. Model Introduction\n\nKimi K2.5 is an open-source, native multimodal agentic model built through continual pretraining on approximately 15 trillion mixed visual and text tokens atop Kimi-K2-Base. It seamlessly integrates vision and language understanding with advanced agentic capabilities, instant and thinking modes, as well as conversational and agentic paradigms.\n\n### Key Features\n- **Native Multimodality**: Pre-trained on vision–language tokens, K2.5 excels in visual knowledge, cross-modal reasoning, and agentic tool use grounded in visual inputs.\n- **Coding with Vision**: K2.5 generates code from visual specifications (UI designs, video workflows) and autonomously orchestrates tools for visual data processing.\n- **Agent Swarm**: K2.5 transitions from single-agent scaling to a self-directed, coordinated swarm-like execution scheme. It decomposes complex tasks into parallel sub-tasks executed by dynamically instantiated, domain-specific agents.\n\n## 2. Model Summary\n\n\u003Cdiv align=\"center\">\n\n\n| | |\n|:---:|:---:|\n| **Architecture** | Mixture-of-Experts (MoE) |\n| **Total Parameters** | 1T |\n| **Activated Parameters** | 32B |\n| **Number of Layers** (Dense layer included) | 61 |\n| **Number of Dense Layers** | 1 |\n| **Attention Hidden Dimension** | 7168 |\n| **MoE Hidden Dimension** (per Expert) | 2048 |\n| **Number of Attention Heads** | 64 |\n| **Number of Experts** | 384 |\n| **Selected Experts per Token** | 8 |\n| **Number of Shared Experts** | 1 |\n| **Vocabulary Size** | 160K |\n| **Context Length** | 256K |\n| **Attention Mechanism** | MLA |\n| **Activation Function** | SwiGLU |\n| **Vision Encoder** | MoonViT |\n| **Parameters of Vision Encoder** | 400M |\n\u003C\u002Fdiv>\n\n## 3. Evaluation Results\n\n\n\n\u003Cdiv align=\"center\">\n\u003Ctable>\n\u003Cthead>\n\u003Ctr>\n\u003Cth align=\"center\">Benchmark\u003C\u002Fth>\n\u003Cth align=\"center\">\u003Csup>Kimi K2.5\u003Cbr>\u003Csup>(Thinking)\u003C\u002Fsup>\u003C\u002Fsup>\u003C\u002Fth>\n\u003Cth align=\"center\">\u003Csup>GPT-5.2 \u003Cbr>\u003Csup>(xhigh)\u003C\u002Fsup>\u003C\u002Fsup>\u003C\u002Fth>\n\u003Cth align=\"center\">\u003Csup>Claude 4.5 Opus \u003Cbr>\u003Csup>(Extended Thinking)\u003C\u002Fsup>\u003C\u002Fsup>\u003C\u002Fth>\n\u003Cth align=\"center\">\u003Csup>Gemini 3 Pro \u003Cbr>\u003Csup>(High Thinking Level)\u003C\u002Fsup>\u003C\u002Fsup>\u003C\u002Fth>\n\u003Cth align=\"center\">\u003Csup>DeepSeek V3.2 \u003Cbr>\u003Csup>(Thinking)\u003C\u002Fsup>\u003C\u002Fsup>\u003C\u002Fth>\n\u003Cth align=\"center\">\u003Csup>Qwen3-VL-\u003Cbr>235B-A22B-\u003Cbr>Thinking\u003C\u002Fsup>\u003C\u002Fth>\n\u003C\u002Ftr>\n\u003C\u002Fthead>\n\u003Ctbody>\n\u003Ctr>\n\u003Ctd align=\"center\" colspan=8>\u003Cstrong>Reasoning & Knowledge\u003C\u002Fstrong>\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">HLE-Full\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">30.1\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">34.5\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">30.8\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">37.5\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">25.1\u003Csup>†\u003C\u002Fsup>\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">HLE-Full\u003Cbr>(w\u002F tools)\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">50.2\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">45.5\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">43.2\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">45.8\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">40.8\u003Csup>†\u003C\u002Fsup>\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">AIME 2025\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">96.1\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">100\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">92.8\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">95.0\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">93.1\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">HMMT 2025 (Feb)\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">95.4\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">99.4\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">92.9*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">97.3*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">92.5\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">IMO-AnswerBench\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">81.8\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">86.3\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">78.5*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">83.1*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">78.3\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">GPQA-Diamond\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">87.6\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">92.4\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">87.0\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">91.9\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">82.4\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">MMLU-Pro\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">87.1\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">86.7*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">89.3*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">90.1\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">85.0\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" colspan=8>\u003Cstrong>Image & Video\u003C\u002Fstrong>\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">MMMU-Pro\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">78.5\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">79.5*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">74.0\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">81.0\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">69.3\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">CharXiv (RQ)\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">77.5\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">82.1\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">67.2*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">81.4\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">66.1\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">MathVision\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">84.2\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">83.0\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">77.1*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">86.1*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">74.6\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">MathVista (mini)\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">90.1\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">82.8*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">80.2*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">89.8*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">85.8\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">ZeroBench\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">9\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">9*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">3*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">8*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">4*\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">ZeroBench\u003Cbr>(w\u002F tools)\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">11\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">7*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">9*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">12*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">3*\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">OCRBench\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">92.3\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">80.7*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">86.5*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">90.3*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">87.5\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">OmniDocBench 1.5\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">88.8\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">85.7\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">87.7*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">88.5\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">82.0*\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">InfoVQA (val)\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">92.6\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">84*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">76.9*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">57.2*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">89.5\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">SimpleVQA\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">71.2\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">55.8*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">69.7*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">69.7*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">56.8*\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">\u003Ca href=\"https:\u002F\u002Fgithub.com\u002FMoonshotAI\u002FWorldVQA\">WorldVQA\u003C\u002Fa>\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">46.3\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">28.0\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">36.8\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">47.4\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">23.5\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">VideoMMMU\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">86.6\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">85.9\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">84.4*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">87.6\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">80.0\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">MMVU\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">80.4\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">80.8*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">77.3\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">77.5\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">71.1\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">MotionBench\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">70.4\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">64.8\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">60.3\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">70.3\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">VideoMME\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">87.4\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">86.0*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">88.4*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">79.0\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">LongVideoBench\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">79.8\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">76.5*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">67.2*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">77.7*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">65.6*\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">LVBench\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">75.9\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">73.5*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">63.6\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" colspan=8>\u003Cstrong>Coding\u003C\u002Fstrong>\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">SWE-Bench Verified\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">76.8\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">80.0\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">80.9\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">76.2\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">73.1\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">SWE-Bench Pro\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">50.7\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">55.6\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">55.4*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">SWE-Bench Multilingual\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">73.0\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">72.0\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">77.5\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">65.0\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">70.2\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">Terminal Bench 2.0\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">50.8\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">54.0\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">59.3\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">54.2\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">46.4\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">PaperBench\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">63.5\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">63.7*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">72.9*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">47.1\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">CyberGym\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">41.3\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">50.6\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">39.9*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">17.3*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">SciCode\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">48.7\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">52.1\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">49.5\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">56.1\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">38.9\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">OJBench (cpp)\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">57.4\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">54.6*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">68.5*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">54.7*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">LiveCodeBench (v6)\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">85.0\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">82.2*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">87.4*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">83.3\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" colspan=8>\u003Cstrong>Long Context\u003C\u002Fstrong>\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">Longbench v2\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">61.0\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">54.5*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">64.4*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">68.2*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">59.8*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">AA-LCR\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">70.0\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">72.3*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">71.3*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">65.3*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">64.3*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctr>\n\u003Ctd align=\"center\" colspan=8>\u003Cstrong>Agentic Search\u003C\u002Fstrong>\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">BrowseComp\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">60.6\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\" rowspan=\"2\">65.8\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">37.0\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">37.8\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">51.4\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">BrowseComp\u003Cbr>(w\u002Fctx manage)\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">74.9\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">57.8\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">59.2\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">67.6\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">BrowseComp\u003Cbr>(Agent Swarm)\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">78.4\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">WideSearch\u003Cbr> (item-f1)\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">72.7\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">76.2*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">57.0\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">32.5*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">WideSearch\u003Cbr> (item-f1 Agent Swarm)\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">79.0\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">DeepSearchQA\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">77.1\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">71.3*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">76.1*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">63.2*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">60.9*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">FinSearchCompT2&T3\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">67.8\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">66.2*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">49.9\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">59.1*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">Seal-0\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">57.4\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">45.0\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">47.7*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">45.5*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">49.5*\u003C\u002Ftd>\n\u003Ctd align=\"center\" style=\"vertical-align: middle\">-\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftbody>\n\u003C\u002Ftable>\n\u003C\u002Fdiv>\n\n\u003Cdetails>\n\u003Csummary>\u003Cb>Footnotes\u003C\u002Fb>\u003C\u002Fsummary>\n\n1. General Testing Details\n   - We report results for Kimi K2.5 and DeepSeek-V3.2 with thinking mode enabled, Claude Opus 4.5 with extended thinking mode, GPT-5.2 with xhigh reasoning effort, and Gemini 3 Pro with a high thinking level. For vision benchmarks, we additionally report results for Qwen3-VL-235B-A22B-Thinking.\n   - Unless otherwise specified, all Kimi K2.5 experiments were conducted with temperature = 1.0, top-p = 0.95, and a context length of 256k tokens.\n   - Benchmarks without publicly available scores were re-evaluated under the same conditions used for Kimi K2.5 and are marked with an asterisk (*).\n   - We could not evaluate GPT-5.2 xhigh on all benchmarks due to service stability issues. For benchmarks that were not tested, we mark them as \"-\".\n2. Text and Reasoning\n   - HLE, AIME 2025, HMMT 2025 (Feb), and GPQA-Diamond were evaluated with a maximum completion budget of 96k tokens.\n   - Results for AIME and HMMT are averaged over 32 runs (avg@32); GPQA-Diamond over 8 runs (avg@8).\n   - For HLE, we report scores on the full set (text & image). Kimi K2.5 scores 31.5 (text) and 21.3 (image) without tools, and 51.8 (text) and 39.8 (image) with tools. The DeepSeek-V3.2 score corresponds to its text-only subset (marked with †) . Hugging Face access was blocked to prevent potential data leakage. HLE with tools uses simple context management: once the context exceeds a threshold, only the latest round of tool messages is retained.\n3. Tool-Augmented \u002F Agentic Search\n   - Kimi K2.5 was equipped with search, code-interpreter, and web-browsing tools for HLE with tools and all agentic search benchmarks.\n   - Except for BrowseComp (where K2.5 and DeepSeek-V3.2 used the discard-all strategy), no context management was applied, and tasks exceeding the supported context length were directly counted as failed.\n   - The test system prompts emphasize deep and proactive tool use, instructing models to reason carefully, leverage tools, and verify uncertain information. Full prompts will be provided in the technical report.\n   - Results for Seal-0 and WideSearch are averaged over four runs (avg@4).\n4. Vision Benchmarks\n   - Max-tokens = 64k, averaged over three runs (avg@3).\n   - ZeroBench (w\u002F tools) uses max-tokens-per-step = 24k and max-steps = 30 for multi-step reasoning.\n   - MMMU-Pro follows the official protocol, preserving input order and prepending images.\n   - GPT-5.2-xhigh had ~10% failure rate (no output despite 3 retries), treated as incorrect; reported scores likely underestimate true performance.\n   - WorldVQA, a benchmark designed to evaluate atomic vision-centric world knowledge. Access WorldVQA at https:\u002F\u002Fgithub.com\u002FMoonshotAI\u002FWorldVQA.\n   - OmniDocBench Score is computed as (1 − normalized Levenshtein distance) × 100, where a higher score denotes superior accuracy.\n5. Coding Tasks\n   - Terminal-Bench 2.0 scores were obtained with the default agent framework (Terminus-2) and the provided JSON parser. In our implementation, we evaluated Terminal-Bench 2.0 under non-thinking mode. This choice was made because our current context management strategy for the thinking mode is incompatible with Terminus-2.\n   - For the SWE-Bench series of evaluations (including verified, multilingual, and pro), we used an internally developed evaluation framework. This framework includes a minimal set of tools—bash tool, createfile tool, insert tool, view tool, strreplace tool, and submit tool—along with tailored system prompts designed for the tasks. The highest scores were achieved under non-thinking mode.\n   - The score of Claude Opus 4.5 on CyberGym is reported under the non-thinking setting.\n   - All reported scores of coding tasks are averaged over 5 independent runs.\n6. Long-Context Benchmarks\n   - AA-LCR: scores averaged over three runs (avg@3).\n   - LongBench-V2: identical prompts and input contexts standardized to ~128k tokens.\n7. Agent Swarm\n   - BrowseComp (Swarm Mode): main agent max 15 steps; sub-agents max 100 steps.\n   - WideSearch (Swarm Mode): main and sub-agents max 100 steps.\n\n\u003C\u002Fdetails>\n\n## 4. Native INT4 Quantization\nKimi-K2.5 adopts the same native int4 quantization method as [Kimi-K2-Thinking](https:\u002F\u002Fhuggingface.co\u002Fmoonshotai\u002FKimi-K2-Thinking#4-native-int4-quantization).\n\n## 5. Deployment\n> [!Note]\n> You can access Kimi-K2.5's API on https:\u002F\u002Fplatform.moonshot.ai and we provide OpenAI\u002FAnthropic-compatible API for you. To verify the deployment is correct, we also provide the  [Kimi Vendor Verifier](https:\u002F\u002Fkimi.com\u002Fblog\u002Fkimi-vendor-verifier.html).\nCurrently, Kimi-K2.5 is recommended to run on the following inference engines:\n* vLLM\n* SGLang\n* KTransformers\n\nThe minimum version requirement for `transformers` is `4.57.1`.\n\nDeployment examples can be found in the [Model Deployment Guide](docs\u002Fdeploy_guidance.md).\n\n\n---\n## 6. Model Usage\n\nThe usage demos below demonstrate how to call our official API.\n\nFor third-party APIs deployed with vLLM or SGLang, please note that:\n> [!Note]\n> - Chat with video content is an experimental feature and is only supported in our official API for now.\n>\n> - The recommended `temperature` will be `1.0` for Thinking mode and `0.6` for Instant mode.\n>\n> - The recommended `top_p` is `0.95`.\n>\n> - To use instant mode, you need to pass `{'chat_template_kwargs': {\"thinking\": False}}` in `extra_body`.\n\n### Chat Completion\n\nThis is a simple chat completion script which shows how to call K2.5 API in Thinking and Instant modes.\n\n```python\nimport openai\nimport base64\nimport requests\ndef simple_chat(client: openai.OpenAI, model_name: str):\n    messages = [\n        {'role': 'system', 'content': 'You are Kimi, an AI assistant created by Moonshot AI.'},\n        {\n            'role': 'user',\n            'content': [\n                {'type': 'text', 'text': 'which one is bigger, 9.11 or 9.9? think carefully.'}\n            ],\n        },\n    ]\n    response = client.chat.completions.create(\n        model=model_name, messages=messages, stream=False, max_tokens=4096\n    )\n    print('====== Below is reasoning_content in Thinking Mode ======')\n    print(f'reasoning content: {response.choices[0].message.reasoning_content}')\n    print('====== Below is response in Thinking Mode ======')\n    print(f'response: {response.choices[0].message.content}')\n\n    # To use instant mode, pass {\"thinking\" = {\"type\":\"disabled\"}}\n    response = client.chat.completions.create(\n        model=model_name,\n        messages=messages,\n        stream=False,\n        max_tokens=4096,\n        extra_body={'thinking': {'type': 'disabled'}},  # this is for official API\n        # extra_body= {'chat_template_kwargs': {\"thinking\": False}}  # this is for vLLM\u002FSGLang\n    )\n    print('====== Below is response in Instant Mode ======')\n    print(f'response: {response.choices[0].message.content}')\n```\n\n\n### Chat Completion with visual content\n\nK2.5 supports Image and Video input.\n\nThe following example demonstrates how to call K2.5 API with image input:\n\n```python\nimport openai\nimport base64\nimport requests\n\ndef chat_with_image(client: openai.OpenAI, model_name: str):\n    url = 'https:\u002F\u002Fhuggingface.co\u002Fmoonshotai\u002FKimi-K2.5\u002Fresolve\u002Fmain\u002Ffigures\u002Fkimi-logo.png'\n    image_base64 = base64.b64encode(requests.get(url).content).decode()\n    messages = [\n        {\n            'role': 'user',\n            'content': [\n                {'type': 'text', 'text': 'Describe this image in detail.'},\n                {\n                    'type': 'image_url',\n                    'image_url': {'url': f'data:image\u002Fpng;base64, {image_base64}'},\n                },\n            ],\n        }\n    ]\n\n    response = client.chat.completions.create(\n        model=model_name, messages=messages, stream=False, max_tokens=8192\n    )\n    print('====== Below is reasoning_content in Thinking Mode ======')\n    print(f'reasoning content: {response.choices[0].message.reasoning_content}')\n    print('====== Below is response in Thinking Mode ======')\n    print(f'response: {response.choices[0].message.content}')\n\n    # Also support instant mode if you pass {\"thinking\" = {\"type\":\"disabled\"}}\n    response = client.chat.completions.create(\n        model=model_name,\n        messages=messages,\n        stream=False,\n        max_tokens=4096,\n        extra_body={'thinking': {'type': 'disabled'}},  # this is for official API\n        # extra_body= {'chat_template_kwargs': {\"thinking\": False}}  # this is for vLLM\u002FSGLang\n    )\n    print('====== Below is response in Instant Mode ======')\n    print(f'response: {response.choices[0].message.content}')\n\n    return response.choices[0].message.content\n```\n\nThe following example demonstrates how to call K2.5 API with video input:\n\n```python\nimport openai\nimport base64\nimport requests\n\ndef chat_with_video(client: openai.OpenAI, model_name:str):\n    url = 'https:\u002F\u002Fhuggingface.co\u002Fmoonshotai\u002FKimi-K2.5\u002Fresolve\u002Fmain\u002Ffigures\u002Fdemo_video.mp4'\n    video_base64 = base64.b64encode(requests.get(url).content).decode()\n    messages = [\n        {\n            \"role\": \"user\",\n            \"content\": [\n                {\"type\": \"text\",\"text\": \"Describe the video in detail.\"},\n                {\n                    \"type\": \"video_url\",\n                    \"video_url\": {\"url\": f\"data:video\u002Fmp4;base64,{video_base64}\"},\n                },\n            ],\n        }\n    ]\n\n    response = client.chat.completions.create(model=model_name, messages=messages)\n    print('====== Below is reasoning_content in Thinking Mode ======')\n    print(f'reasoning content: {response.choices[0].message.reasoning_content}')\n    print('====== Below is response in Thinking Mode ======')\n    print(f'response: {response.choices[0].message.content}')\n\n    # Also support instant mode if pass {\"thinking\" = {\"type\":\"disabled\"}}\n    response = client.chat.completions.create(\n        model=model_name,\n        messages=messages,\n        stream=False,\n        max_tokens=4096,\n        extra_body={'thinking': {'type': 'disabled'}},  # this is for official API\n        # extra_body= {'chat_template_kwargs': {\"thinking\": False}}  # this is for vLLM\u002FSGLang\n    )\n    print('====== Below is response in Instant Mode ======')\n    print(f'response: {response.choices[0].message.content}')\n    return response.choices[0].message.content\n```\n\n### Interleaved Thinking and Multi-Step Tool Call\n\nK2.5 shares the same design of Interleaved Thinking and Multi-Step Tool Call as K2 Thinking. For usage example, please refer to the [K2 Thinking documentation](https:\u002F\u002Fplatform.moonshot.ai\u002Fdocs\u002Fguide\u002Fuse-kimi-k2-thinking-model#complete-example).\n\n\n### Coding Agent Framework\n\nKimi K2.5 works best with Kimi Code CLI as its agent framework — give it a try at https:\u002F\u002Fwww.kimi.com\u002Fcode.\n\n\n---\n\n## 7. License\n\nBoth the code repository and the model weights are released under the [Modified MIT License](LICENSE).\n\n\n---\n\n## 9. Contact Us\n\nIf you have any questions, please reach out at [support@moonshot.cn](mailto:support@moonshot.cn).\n","Kimi K2.5 是 Moonshot AI 开发的一款强大的多模态智能代理模型。该项目通过在约15万亿个混合视觉和文本令牌上进行持续预训练，实现了视觉与语言理解的无缝集成，并具备高级代理功能、即时和思考模式以及对话和代理范式。其核心技术特点包括原生多模态能力，能够从视觉规范生成代码并自主协调工具处理视觉数据；同时，Kimi K2.5 采用了一种自指导、协调式的群体执行方案，可以将复杂任务分解为多个并行子任务，由动态实例化的领域特定代理执行。该模型适用于需要高度集成视觉理解和自然语言处理的应用场景，如自动编码、视觉数据分析等。",2,"2026-06-11 03:52:28","high_star"]