| <!DOCTYPE html> |
| <html lang="en"> |
| <head> |
| <meta charset="UTF-8"> |
| <meta name="viewport" content="width=device-width, initial-scale=1.0"> |
| <title>KIMI K2: Open Agentic Intelligence</title> |
| <link rel="stylesheet" href="https://cdnjs.cloudflare.com/ajax/libs/font-awesome/6.4.0/css/all.min.css"> |
| <style> |
| :root { |
| --primary: #2563eb; |
| --secondary: #1e40af; |
| --accent: #3b82f6; |
| --dark: #1e1b4b; |
| --light: #f8fafc; |
| --text: #1f2937; |
| --highlight: #10b981; |
| } |
| |
| * { |
| margin: 0; |
| padding: 0; |
| box-sizing: border-box; |
| font-family: 'Segoe UI', Tahoma, Geneva, Verdana, sans-serif; |
| } |
| |
| body { |
| background-color: var(--dark); |
| color: var(--light); |
| line-height: 1.6; |
| } |
| |
| .poster { |
| max-width: 1200px; |
| margin: 2rem auto; |
| padding: 2rem; |
| background: linear-gradient(135deg, var(--dark) 0%, #1e1b4b 100%); |
| border-radius: 16px; |
| box-shadow: 0 25px 50px -12px rgba(0, 0, 0, 0.5); |
| position: relative; |
| overflow: hidden; |
| border: 1px solid rgba(255, 255, 255, 0.1); |
| } |
| |
| .poster::before { |
| content: ''; |
| position: absolute; |
| top: 0; |
| left: 0; |
| right: 0; |
| height: 4px; |
| background: linear-gradient(90deg, var(--primary), var(--highlight), var(--accent)); |
| } |
| |
| .header { |
| text-align: center; |
| margin-bottom: 2rem; |
| position: relative; |
| z-index: 1; |
| } |
| |
| .title { |
| font-size: 3.5rem; |
| font-weight: 800; |
| margin-bottom: 1rem; |
| background: linear-gradient(90deg, var(--primary), var(--accent)); |
| -webkit-background-clip: text; |
| background-clip: text; |
| color: transparent; |
| text-transform: uppercase; |
| letter-spacing: 1px; |
| } |
| |
| .subtitle { |
| font-size: 1.5rem; |
| font-weight: 600; |
| color: var(--accent); |
| margin-bottom: 0.5rem; |
| } |
| |
| .team { |
| font-size: 1.2rem; |
| color: var(--light); |
| opacity: 0.8; |
| margin-bottom: 2rem; |
| } |
| |
| .content { |
| display: grid; |
| grid-template-columns: 1fr 1fr; |
| gap: 2rem; |
| margin-bottom: 2rem; |
| } |
| |
| .abstract { |
| background: rgba(30, 64, 175, 0.2); |
| padding: 1.5rem; |
| border-radius: 12px; |
| border-left: 4px solid var(--primary); |
| backdrop-filter: blur(10px); |
| } |
| |
| .abstract-title { |
| font-size: 1.5rem; |
| font-weight: 700; |
| margin-bottom: 1rem; |
| color: var(--highlight); |
| } |
| |
| .abstract-text { |
| margin-bottom: 1rem; |
| font-size: 1.1rem; |
| } |
| |
| .highlights { |
| background: rgba(16, 185, 129, 0.1); |
| padding: 1.5rem; |
| border-radius: 12px; |
| border-left: 4px solid var(--highlight); |
| backdrop-filter: blur(10px); |
| } |
| |
| .highlights-title { |
| font-size: 1.5rem; |
| font-weight: 700; |
| margin-bottom: 1rem; |
| color: var(--highlight); |
| } |
| |
| .highlight-item { |
| display: flex; |
| align-items: flex-start; |
| margin-bottom: 1rem; |
| } |
| |
| .highlight-icon { |
| color: var(--highlight); |
| margin-right: 1rem; |
| font-size: 1.2rem; |
| margin-top: 0.2rem; |
| } |
| |
| .stats { |
| display: grid; |
| grid-template-columns: repeat(2, 1fr); |
| gap: 1rem; |
| margin-top: 2rem; |
| } |
| |
| .stat-item { |
| background: rgba(59, 130, 246, 0.1); |
| padding: 1rem; |
| border-radius: 8px; |
| text-align: center; |
| border: 1px solid rgba(59, 130, 246, 0.3); |
| } |
| |
| .stat-value { |
| font-size: 2rem; |
| font-weight: 700; |
| color: var(--accent); |
| margin-bottom: 0.5rem; |
| } |
| |
| .stat-label { |
| font-size: 0.9rem; |
| opacity: 0.8; |
| } |
| |
| .image-container { |
| margin: 2rem 0; |
| text-align: center; |
| } |
| |
| .image-container img { |
| max-width: 100%; |
| border-radius: 8px; |
| box-shadow: 0 10px 15px -3px rgba(0, 0, 0, 0.3); |
| border: 1px solid rgba(255, 255, 255, 0.1); |
| } |
| |
| .footer { |
| text-align: center; |
| margin-top: 2rem; |
| padding-top: 1rem; |
| border-top: 1px solid rgba(255, 255, 255, 0.1); |
| font-size: 0.9rem; |
| opacity: 0.7; |
| } |
| |
| .footer a { |
| color: var(--accent); |
| text-decoration: none; |
| transition: color 0.3s; |
| } |
| |
| .footer a:hover { |
| color: var(--highlight); |
| } |
| |
| .tech-grid { |
| display: grid; |
| grid-template-columns: repeat(3, 1fr); |
| gap: 1rem; |
| margin-top: 2rem; |
| } |
| |
| .tech-item { |
| background: rgba(30, 64, 175, 0.2); |
| padding: 1rem; |
| border-radius: 8px; |
| text-align: center; |
| transition: transform 0.3s, box-shadow 0.3s; |
| } |
| |
| .tech-item:hover { |
| transform: translateY(-5px); |
| box-shadow: 0 10px 20px -5px rgba(0, 0, 0, 0.2); |
| } |
| |
| .tech-icon { |
| font-size: 2rem; |
| color: var(--accent); |
| margin-bottom: 0.5rem; |
| } |
| |
| .tech-title { |
| font-weight: 600; |
| margin-bottom: 0.5rem; |
| } |
| |
| .tech-desc { |
| font-size: 0.9rem; |
| opacity: 0.8; |
| } |
| |
| @media (max-width: 768px) { |
| .content { |
| grid-template-columns: 1fr; |
| } |
| |
| .title { |
| font-size: 2.5rem; |
| } |
| |
| .stats, .tech-grid { |
| grid-template-columns: 1fr; |
| } |
| } |
| </style> |
| </head> |
| <body> |
| <div class="poster"> |
| <div class="header"> |
| <h1 class="title">KIMI K2</h1> |
| <h2 class="subtitle">Open Agentic Intelligence</h2> |
| <p class="team">Technical Report · Kimi Team</p> |
| </div> |
|
|
| <div class="content"> |
| <div class="abstract"> |
| <h3 class="abstract-title">ABSTRACT</h3> |
| <p class="abstract-text">We introduce Kimi K2, a Mixture-of-Experts (MoE) large language model with 32 billion activated parameters and 1 trillion total parameters. We propose the MuonClip optimizer, which improves upon Muon with a novel QK-clip technique to address training instability while enjoying the advanced token efficiency of Muon.</p> |
| <p class="abstract-text">Kimi K2 achieves state-of-the-art performance among open-source non-thinking models, with strengths in agentic capabilities. Notably, K2 obtains 66.1 on Tau2-Bench, 76.5 on ACEBench (En), 65.8 on SWE-Bench Verified, and 47.3 on SWE-Bench Multilingual — surpassing most open and closed-sourced baselines in non-thinking settings.</p> |
| </div> |
|
|
| <div class="highlights"> |
| <h3 class="highlights-title">KEY HIGHLIGHTS</h3> |
| <div class="highlight-item"> |
| <div class="highlight-icon"><i class="fas fa-brain"></i></div> |
| <div> |
| <strong>1.04 Trillion Parameters</strong> - MoE architecture with 32B activated parameters |
| </div> |
| </div> |
| <div class="highlight-item"> |
| <div class="highlight-icon"><i class="fas fa-chart-line"></i></div> |
| <div> |
| <strong>15.5 Trillion Tokens</strong> - Pre-trained with zero loss spikes |
| </div> |
| </div> |
| <div class="highlight-item"> |
| <div class="highlight-icon"><i class="fas fa-robot"></i></div> |
| <div> |
| <strong>Agentic Capabilities</strong> - Superior performance in software engineering and agentic tasks |
| </div> |
| </div> |
| </div> |
| </div> |
|
|
| <div class="stats"> |
| <div class="stat-item"> |
| <div class="stat-value">66.1</div> |
| <div class="stat-label">Tau2-Bench</div> |
| </div> |
| <div class="stat-item"> |
| <div class="stat-value">76.5</div> |
| <div class="stat-label">ACEBench (En)</div> |
| </div> |
| <div class="stat-item"> |
| <div class="stat-value">65.8</div> |
| <div class="stat-label">SWE-Bench Verified</div> |
| </div> |
| <div class="stat-item"> |
| <div class="stat-value">47.3</div> |
| <div class="stat-label">SWE-Bench Multilingual</div> |
| </div> |
| </div> |
|
|
| <div class="image-container"> |
| <img src="https://cdn.vansin.top/picgo/2301e0231d5e471d6d6231ca53eaccfb06089892a33c2d2311d5805ee440c93a.jpg" alt="Kimi K2 Results"> |
| </div> |
|
|
| <div class="tech-grid"> |
| <div class="tech-item"> |
| <div class="tech-icon"><i class="fas fa-cogs"></i></div> |
| <h4 class="tech-title">MuonClip Optimizer</h4> |
| <p class="tech-desc">Novel QK-clip technique for stable training</p> |
| </div> |
| <div class="tech-item"> |
| <div class="tech-icon"><i class="fas fa-database"></i></div> |
| <h4 class="tech-title">Agentic Data Pipeline</h4> |
| <p class="tech-desc">Large-scale synthesis of tool-use demonstrations</p> |
| </div> |
| <div class="tech-item"> |
| <div class="tech-icon"><i class="fas fa-robot"></i></div> |
| <h4 class="tech-title">RL Framework</h4> |
| <p class="tech-desc">Combines verifiable rewards with self-critique</p> |
| </div> |
| </div> |
|
|
| <div class="footer"> |
| Created by <a href="https://huggingface.co/spaces/akhaliq/anycoder" target="_blank">AnyCoder</a> |
| </div> |
| </div> |
| </body> |
| </html> |
