openbmb/MiniCPM5-2B-Midtrain
224k
1---2license: apache-2.03language:4 - en5 - zh6library_name: transformers7pipeline_tag: text-generation8tags:9 - minicpm10 - minicpm511 - llama12 - text-generation13 - long-context14 - tool-calling15 - on-device16 - edge-ai17datasets:18 - openbmb/Ultra-FineWeb19 - openbmb/UltraX-Preview20 - openbmb/Ultra-FineWeb-L321 - openbmb/UltraData-Math22 - openbmb/UltraData-Code23 - openbmb/UltraData-SFT-260524 - openbmb/UltraData-SFT-Agent-260925 - openbmb/UltraData-RL-260926---27 28<div align="center">29<img src="https://raw.githubusercontent.com/OpenBMB/MiniCPM/main/assets/minicpm_logo.png" width="500em" />30</div>31 32<p align="center">33<a href="https://arxiv.org/pdf/2506.07900" target="_blank">MiniCPM Tech Report</a> |34<a href="https://modelbest.feishu.cn/wiki/UtWxwcERfiRIpIkBOjuc3h9tn1D" target="_blank">MiniCPM Wiki(Chinese)</a> |35<a href="https://github.com/OpenBMB/MiniCPM" target="_blank">GitHub Repo</a> |36<a href="https://ultradata.openbmb.cn/" target="_blank">UltraData</a> |37<a href="https://huggingface.co/spaces/openbmb/MiniCPM5-2B-Demo" target="_blank">Online Demo</a>38</p>39 40<p align="center">41English |42<a href="https://huggingface.co/openbmb/MiniCPM5-2B/blob/main/README-cn.md" target="_blank">中文</a>43</p>44 45## Highlights46 47We are releasing **MiniCPM5-2B**, the second model in the **MiniCPM5** series, following [MiniCPM5-1B](https://huggingface.co/openbmb/MiniCPM5-1B). It is a dense 2B Transformer that scales up the same training recipe, built for on-device, local deployment, and resource-constrained scenarios, reaching 2B-class open-source SOTA.48 49🏆 **2B-class open-source SOTA**: compared with strong open-source models of similar size, MiniCPM5-2B achieves SOTA performance within this comparison set. It remains competitive with 4B-class models overall, while showing its advantages over models of comparable size in coding, mathematics, long-context understanding, tool use, and agentic tasks.50 51<div id="capability-comparison-radar" class="radar-visual" role="img" aria-label="Capability radar chart comparing MiniCPM5-2B with 4B-class models. Each axis is normalized independently to 100 percent.">52 <style>53 #capability-comparison-radar {54 --foreground: #171717;55 display: block; width: 100%; max-width: 620px; margin: 0 auto; background: #fff;56 font-family: -apple-system, BlinkMacSystemFont, "Segoe UI", Roboto, "PingFang SC", "Hiragino Sans GB", "Microsoft YaHei", sans-serif;57 }58 #capability-comparison-radar svg { display: block; width: 100%; height: auto; background: #fff; }59 #capability-comparison-radar .title { fill: var(--foreground); font-size: 16px; font-weight: 600; letter-spacing: 0; }60 #capability-comparison-radar .axis-label { fill: #333; font-size: 12px; font-weight: 600; }61 #capability-comparison-radar .ring { fill: none; stroke: rgba(128,128,128,.18); stroke-width: .8; }62 #capability-comparison-radar .spoke { stroke: rgba(128,128,128,.25); stroke-width: .8; }63 #capability-comparison-radar .ring-label { fill: #8a8a8a; font-size: 9px; }64 #capability-comparison-radar .series { stroke-linejoin: round; }65 #capability-comparison-radar .series.primary { stroke-width: 2.4; }66 #capability-comparison-radar .legend-label { fill: #171717; font-size: 12px; font-weight: 600; }67 #capability-comparison-radar .legend-average { fill: #666; font-size: 11px; }68 #capability-comparison-radar .legend-swatch { rx: 3; }69 @media (max-width: 900px) { #capability-comparison-radar { overflow-x: auto; } #capability-comparison-radar svg { min-width: 620px; } }70 </style>71 <svg viewBox="0 0 680 560" aria-hidden="true">72 <g transform="translate(-20 0)">73 <text x="40" y="28" class="title" text-anchor="start">Capability Radar by Dimension</text>74 <polygon points="340.0,235.0 362.5,243.2 374.5,263.9 370.3,287.5 352.0,302.9 328.0,302.9 309.7,287.5 305.5,263.9 317.5,243.2" class="ring" stroke="rgba(128,128,128,.18)" stroke-width=".8"/>75 <text x="344.0" y="238.0" class="ring-label">20%</text>76 <polygon points="340.0,200.0 385.0,216.4 408.9,257.8 400.6,305.0 363.9,335.8 316.1,335.8 279.4,305.0 271.1,257.8 295.0,216.4" class="ring" stroke="rgba(128,128,128,.18)" stroke-width=".8"/>77 <text x="344.0" y="203.0" class="ring-label">40%</text>78 <polygon points="340.0,165.0 407.5,189.6 443.4,251.8 430.9,322.5 375.9,368.7 304.1,368.7 249.1,322.5 236.6,251.8 272.5,189.6" class="ring" stroke="rgba(128,128,128,.18)" stroke-width=".8"/>79 <text x="344.0" y="168.0" class="ring-label">60%</text>80 <polygon points="340.0,130.0 430.0,162.8 477.9,245.7 461.2,340.0 387.9,401.6 292.1,401.6 218.8,340.0 202.1,245.7 250.0,162.8" class="ring" stroke="rgba(128,128,128,.18)" stroke-width=".8"/>81 <text x="344.0" y="133.0" class="ring-label">80%</text>82 <polygon points="340.0,95.0 452.5,135.9 512.3,239.6 491.6,357.5 399.9,434.4 280.1,434.4 188.4,357.5 167.7,239.6 227.5,135.9" class="ring" stroke="rgba(29,111,208,.28)" stroke-width="1"/>83 <text x="344.0" y="98.0" class="ring-label">100%</text>84 <line x1="340" y1="270" x2="340.0" y2="95.0" class="spoke"/>85 <text x="340.0" y="63.5" class="axis-label" text-anchor="middle">Code Reasoning</text>86 <line x1="340" y1="270" x2="452.5" y2="135.9" class="spoke"/>87 <text x="472.7" y="111.8" class="axis-label" text-anchor="start">Math Reasoning</text>88 <line x1="340" y1="270" x2="512.3" y2="239.6" class="spoke"/>89 <text x="543.4" y="234.1" class="axis-label" text-anchor="start">Instruction Following</text>90 <line x1="340" y1="270" x2="491.6" y2="357.5" class="spoke"/>91 <text x="518.8" y="373.2" class="axis-label" text-anchor="start">General Knowledge</text>92 <line x1="340" y1="270" x2="399.9" y2="434.4" class="spoke"/>93 <text x="410.6" y="464.0" class="axis-label" text-anchor="start">Long Context</text>94 <line x1="340" y1="270" x2="280.1" y2="434.4" class="spoke"/>95 <text x="269.4" y="464.0" class="axis-label" text-anchor="end">Tool Use</text>96 <line x1="340" y1="270" x2="188.4" y2="357.5" class="spoke"/>97 <text x="161.2" y="373.3" class="axis-label" text-anchor="end">Coding Agent</text>98 <line x1="340" y1="270" x2="167.7" y2="239.6" class="spoke"/>99 <text x="136.6" y="234.1" class="axis-label" text-anchor="end">Search Agent</text>100 <line x1="340" y1="270" x2="227.5" y2="135.9" class="spoke"/>101 <text x="207.3" y="111.8" class="axis-label" text-anchor="end">General Agent</text>102 <polygon points="340.0,136.7 450.1,138.8 498.4,242.1 491.6,357.5 398.4,430.3 289.5,408.8 188.4,357.5 188.2,243.2 249.3,161.9" class="series" fill="#E85D4C" fill-opacity=".10" stroke="#E85D4C" stroke-width="1.5"/>103 <circle cx="340.0" cy="136.7" r="3" fill="#E85D4C"/>104 <circle cx="450.1" cy="138.8" r="3" fill="#E85D4C"/>105 <circle cx="498.4" cy="242.1" r="3" fill="#E85D4C"/>106 <circle cx="491.6" cy="357.5" r="3" fill="#E85D4C"/>107 <circle cx="398.4" cy="430.3" r="3" fill="#E85D4C"/>108 <circle cx="289.5" cy="408.8" r="3" fill="#E85D4C"/>109 <circle cx="188.4" cy="357.5" r="3" fill="#E85D4C"/>110 <circle cx="188.2" cy="243.2" r="3" fill="#E85D4C"/>111 <circle cx="249.3" cy="161.9" r="3" fill="#E85D4C"/>112 <polygon points="340.0,134.4 448.9,140.2 512.3,239.6 462.2,340.5 367.8,346.5 308.0,358.0 231.0,332.9 242.3,252.8 250.1,162.9" class="series" fill="#2A9D8F" fill-opacity=".10" stroke="#2A9D8F" stroke-width="1.5"/>113 <circle cx="340.0" cy="134.4" r="3" fill="#2A9D8F"/>114 <circle cx="448.9" cy="140.2" r="3" fill="#2A9D8F"/>115 <circle cx="512.3" cy="239.6" r="3" fill="#2A9D8F"/>116 <circle cx="462.2" cy="340.5" r="3" fill="#2A9D8F"/>117 <circle cx="367.8" cy="346.5" r="3" fill="#2A9D8F"/>118 <circle cx="308.0" cy="358.0" r="3" fill="#2A9D8F"/>119 <circle cx="231.0" cy="332.9" r="3" fill="#2A9D8F"/>120 <circle cx="242.3" cy="252.8" r="3" fill="#2A9D8F"/>121 <circle cx="250.1" cy="162.9" r="3" fill="#2A9D8F"/>122 <polygon points="340.0,189.3 411.4,184.9 502.9,241.3 455.3,336.6 356.7,315.9 288.5,411.4 320.8,281.1 266.8,257.1 298.8,220.9" class="series" fill="#E09F3E" fill-opacity=".10" stroke="#E09F3E" stroke-width="1.5"/>123 <circle cx="340.0" cy="189.3" r="3" fill="#E09F3E"/>124 <circle cx="411.4" cy="184.9" r="3" fill="#E09F3E"/>125 <circle cx="502.9" cy="241.3" r="3" fill="#E09F3E"/>126 <circle cx="455.3" cy="336.6" r="3" fill="#E09F3E"/>127 <circle cx="356.7" cy="315.9" r="3" fill="#E09F3E"/>128 <circle cx="288.5" cy="411.4" r="3" fill="#E09F3E"/>129 <circle cx="320.8" cy="281.1" r="3" fill="#E09F3E"/>130 <circle cx="266.8" cy="257.1" r="3" fill="#E09F3E"/>131 <circle cx="298.8" cy="220.9" r="3" fill="#E09F3E"/>132 <polygon points="340.0,95.0 452.5,135.9 499.8,241.8 476.2,348.7 399.9,434.4 280.1,434.4 220.0,339.3 167.7,239.6 227.5,135.9" class="series primary" fill="#1D6FD0" fill-opacity=".22" stroke="#1D6FD0" stroke-width="2.4"/>133 <circle cx="340.0" cy="95.0" r="3" fill="#1D6FD0"/>134 <circle cx="452.5" cy="135.9" r="3" fill="#1D6FD0"/>135 <circle cx="499.8" cy="241.8" r="3" fill="#1D6FD0"/>136 <circle cx="476.2" cy="348.7" r="3" fill="#1D6FD0"/>137 <circle cx="399.9" cy="434.4" r="3" fill="#1D6FD0"/>138 <circle cx="280.1" cy="434.4" r="3" fill="#1D6FD0"/>139 <circle cx="220.0" cy="339.3" r="3" fill="#1D6FD0"/>140 <circle cx="167.7" cy="239.6" r="3" fill="#1D6FD0"/>141 <circle cx="227.5" cy="135.9" r="3" fill="#1D6FD0"/>142 <rect x="40" y="490" width="18" height="18" class="legend-swatch" fill="#1D6FD0"/>143 <text x="66" y="504" class="legend-label">MiniCPM5-2B</text>144 <text x="66" y="520" class="legend-average">avg 53.9</text>145 <rect x="195" y="490" width="18" height="18" class="legend-swatch" fill="#E85D4C"/>146 <text x="221" y="504" class="legend-label">Qwen3.5-4B</text>147 <text x="221" y="520" class="legend-average">avg 51.1</text>148 <rect x="350" y="490" width="18" height="18" class="legend-swatch" fill="#2A9D8F"/>149 <text x="376" y="504" class="legend-label">granite-4.2-3B</text>150 <text x="376" y="520" class="legend-average">avg 42.7</text>151 <rect x="505" y="490" width="18" height="18" class="legend-swatch" fill="#E09F3E"/>152 <text x="531" y="504" class="legend-label">LFM2.5-2.6B</text>153 <text x="531" y="520" class="legend-average">avg 33.2</text>154 <text x="340" y="548" class="legend-average" text-anchor="middle">each axis: max = 100%</text>155 </g>156 </svg>157</div>158 159📂 **Open High-Quality Data**: Alongside the model, we are releasing the high-quality training datasets behind it as part of the [UltraData](https://ultradata.openbmb.cn/) family: [UltraX](https://huggingface.co/datasets/openbmb/UltraX-Preview), a high-quality web pre-training dataset; [UltraData-Code](https://huggingface.co/datasets/openbmb/UltraData-Code), featuring L0–L3 tiered code data management to drive a significant leap in coding capabilities; [UltraData-SFT-Agent-2609](https://huggingface.co/datasets/openbmb/UltraData-SFT-Agent-2609), comprising 500K agent training samples to enhance comprehensive on-device agent capabilities; and [UltraData-RL-2609](https://huggingface.co/datasets/openbmb/UltraData-RL-2609), with 80K+ high-quality RL training samples covering mathematics, code, general knowledge, and long-context reasoning.160 161## Model List162 163Use this directory to choose the model format that matches your runtime:164 165**MiniCPM5-2B**166 167- **[MiniCPM5-2B](https://huggingface.co/openbmb/MiniCPM5-2B)** · [ModelScope](https://www.modelscope.cn/models/OpenBMB/MiniCPM5-2B) · BF16 final release (post-trained with RL + OPD)168- **[MiniCPM5-2B-SFT](https://huggingface.co/openbmb/MiniCPM5-2B-SFT)** · [ModelScope](https://www.modelscope.cn/models/OpenBMB/MiniCPM5-2B-SFT) · BF16 SFT-only checkpoint (before RL / OPD)169- **[MiniCPM5-2B-Midtrain](https://huggingface.co/openbmb/MiniCPM5-2B-Midtrain)** · [ModelScope](https://www.modelscope.cn/models/OpenBMB/MiniCPM5-2B-Midtrain) · BF16 mid-training checkpoint (before SFT) **👈 you are here**170- **[MiniCPM5-2B-Base](https://huggingface.co/openbmb/MiniCPM5-2B-Base)** · [ModelScope](https://www.modelscope.cn/models/OpenBMB/MiniCPM5-2B-Base) · BF16 base checkpoint (pre-training only)171- **[MiniCPM5-2B-GGUF](https://huggingface.co/openbmb/MiniCPM5-2B-GGUF)** · [ModelScope](https://www.modelscope.cn/models/OpenBMB/MiniCPM5-2B-GGUF) · GGUF for llama.cpp / Ollama / LM Studio172- **[MiniCPM5-2B-MLX](https://huggingface.co/openbmb/MiniCPM5-2B-MLX)** · [ModelScope](https://www.modelscope.cn/models/OpenBMB/MiniCPM5-2B-MLX) · MLX / 4bit for Apple Silicon173- **[MiniCPM5-2B-GPTQ](https://huggingface.co/openbmb/MiniCPM5-2B-GPTQ)** · [ModelScope](https://www.modelscope.cn/models/OpenBMB/MiniCPM5-2B-GPTQ) · GPTQ / 4bit quantized model174- **[MiniCPM5-2B-DSpark](https://huggingface.co/openbmb/MiniCPM5-2B-DSpark)** · [ModelScope](https://www.modelscope.cn/models/OpenBMB/MiniCPM5-2B-DSpark) · DSpark draft model for inference acceleration175- **[MiniCPM5-2B-DSpark-GGUF](https://huggingface.co/openbmb/MiniCPM5-2B-DSpark-GGUF)** · [ModelScope](https://www.modelscope.cn/models/OpenBMB/MiniCPM5-2B-DSpark-GGUF) · GGUF version of DSpark draft model176- **[MiniCPM5-2B-LiteRT](https://huggingface.co/litert-community/MiniCPM5-2B)** · [ModelScope](https://www.modelscope.cn/models/litert-community/MiniCPM5-2B) · the LiteRT-LM version of MiniCPM5-2B177 178**MiniCPM5-1B**179 180- **[MiniCPM5-1B](https://huggingface.co/openbmb/MiniCPM5-1B)** · [ModelScope](https://www.modelscope.cn/models/OpenBMB/MiniCPM5-1B) · BF16 final release (post-trained with RL + OPD)181- **[MiniCPM5-1B-SFT](https://huggingface.co/openbmb/MiniCPM5-1B-SFT)** · [ModelScope](https://www.modelscope.cn/models/OpenBMB/MiniCPM5-1B-SFT) · BF16 SFT-only checkpoint (before RL / OPD)182- **[MiniCPM5-1B-Base](https://huggingface.co/openbmb/MiniCPM5-1B-Base)** · [ModelScope](https://www.modelscope.cn/models/OpenBMB/MiniCPM5-1B-Base) · BF16 base checkpoint (pre-training only)183- **[MiniCPM5-1B-GGUF](https://huggingface.co/openbmb/MiniCPM5-1B-GGUF)** · [ModelScope](https://www.modelscope.cn/models/OpenBMB/MiniCPM5-1B-GGUF) · GGUF for llama.cpp / Ollama / LM Studio184- **[MiniCPM5-1B-MLX](https://huggingface.co/openbmb/MiniCPM5-1B-MLX)** · [ModelScope](https://www.modelscope.cn/models/OpenBMB/MiniCPM5-1B-MLX) · MLX / 4bit for Apple Silicon185 186## Model Information187 188MiniCPM5-2B has the following features:189 190- **Type**: Causal Language Model191- **Architecture**: Standard `LlamaForCausalLM`192- **Number of Parameters**: 2,516,756,480193- **Number of Non-Embedding Parameters**: 1,981,982,720194- **Number of Layers**: 42195- **Number of Attention Heads (GQA)**: 16 for Q and 2 for KV196- **Context Length**: 131,072197 198## Introduction199 200MiniCPM5-2B is the second model in the MiniCPM5 series. It is designed for local assistants, coding agents, tool-use workflows, and reasoning scenarios where a compact model is preferred. The model keeps a small deployment footprint while providing native long-context support.201 202## Evaluation Results203 204We compare **MiniCPM5-2B** with strong open-source models in the same size class, including **LFM2.5-2.6B**, **Qwen3.5-2B**, and **Gemma-4-E2B-it**, while also listing larger models such as **Qwen3.5-4B**, **granite-4.2-3B**, **Nemotron-3-Nano-4B**, **Gemma-4-E4B-it**, and **LFM2.5-8B-A1B** for reference.205 206Within this comparison set, MiniCPM5-2B reaches 2B-class open-source SOTA with an average score of **53.9**, and also exceeds all of the larger models included here (the highest is **51.1**). Its advantages are most visible in code reasoning, math reasoning, long-context understanding, tool use, and multiple agentic tasks.207 208<div style="width:100%;max-width:1080px;margin:0 auto;padding:16px 0;background:#fff;209font-family:-apple-system,BlinkMacSystemFont,'Segoe UI',Roboto,'PingFang SC',210'Hiragino Sans GB','Microsoft YaHei',sans-serif;color:#171717">211 <h1 style="margin:0 0 14px;font-size:22px;font-weight:700;color:#1D6FD0;212 letter-spacing:0.02em">Evaluation Results of MiniCPM5-2B and Baselines</h1>213 <table class="vl-table" style="width:100%;margin:0;table-layout:fixed;border-collapse:collapse;font-size:13px;font-variant-numeric:tabular-nums"><thead><tr><th rowspan="2" style="padding:7px 5px;text-align:left;font-weight:600;border-bottom:2px solid #1D6FD0;color:#1D6FD0;width:18%"></th><th rowspan="2" style="padding:7px 4px;text-align:center;font-weight:600;border-bottom:2px solid #1D6FD0;color:#1D6FD0;font-size:12px;width:9.111%;background:rgba(29, 111, 208, 0.08);vertical-align:middle;word-break:normal;">MiniCPM5-2B</th><th colspan="3" style="padding:6px 4px;text-align:center;font-weight:600;color:#1D6FD0;font-size:13px;border-bottom:1px solid rgba(29, 111, 208, 0.2);border-left:1px solid rgba(29, 111, 208, 0.25);">2B-class Models</th><th colspan="5" style="padding:6px 4px;text-align:center;font-weight:600;color:#1D6FD0;font-size:13px;border-bottom:1px solid rgba(29, 111, 208, 0.2);border-left:1px solid rgba(29, 111, 208, 0.25);">4B-class Models</th></tr><tr><th style="padding:6px 4px;text-align:center;font-weight:500;border-bottom:2px solid #1D6FD0;color:#1D6FD0;font-size:11px;width:9.111%;border-left:1px solid rgba(29, 111, 208, 0.25);word-break:normal;vertical-align:middle;">LFM2.5-2.6B</th><th style="padding:6px 4px;text-align:center;font-weight:500;border-bottom:2px solid #1D6FD0;color:#1D6FD0;font-size:11px;width:9.111%;word-break:normal;vertical-align:middle;">Qwen3.5-2B</th><th style="padding:6px 4px;text-align:center;font-weight:500;border-bottom:2px solid #1D6FD0;color:#1D6FD0;font-size:11px;width:9.111%;word-break:normal;vertical-align:middle;">Gemma-4-E2B-it</th><th style="padding:6px 4px;text-align:center;font-weight:500;border-bottom:2px solid #1D6FD0;color:#1D6FD0;font-size:11px;width:9.111%;border-left:1px solid rgba(29, 111, 208, 0.25);word-break:normal;vertical-align:middle;">Qwen3.5-4B</th><th style="padding:6px 4px;text-align:center;font-weight:500;border-bottom:2px solid #1D6FD0;color:#1D6FD0;font-size:11px;width:9.111%;word-break:normal;vertical-align:middle;">granite-4.2-3B</th><th style="padding:6px 4px;text-align:center;font-weight:500;border-bottom:2px solid #1D6FD0;color:#1D6FD0;font-size:11px;width:9.111%;word-break:normal;vertical-align:middle;">Nemotron-3-Nano-4B</th><th style="padding:6px 4px;text-align:center;font-weight:500;border-bottom:2px solid #1D6FD0;color:#1D6FD0;font-size:11px;width:9.111%;word-break:normal;vertical-align:middle;">Gemma-4-E4B-it</th><th style="padding:6px 4px;text-align:center;font-weight:500;border-bottom:2px solid #1D6FD0;color:#1D6FD0;font-size:11px;width:9.111%;word-break:normal;vertical-align:middle;">LFM2.5-8B-A1B</th></tr></thead><tbody>214<tr style="background:rgba(29, 111, 208, 0.03)"><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">Average</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;font-weight:600;"><strong style="color:#1D6FD0">53.9</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;font-weight:600;">33.2</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;font-weight:600;">28.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;font-weight:600;">24.6</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;font-weight:600;">51.1</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;font-weight:600;">42.7</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;font-weight:600;">32.6</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;font-weight:600;">31.2</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;font-weight:600;">28.4</td></tr>215<tr><td colspan="10" style="padding:5px 10px;font-weight:600;color:#1D6FD0;border-bottom:1px solid rgba(29, 111, 208, 0.2);background:rgba(29, 111, 208, 0.14)">Code Reasoning</td></tr>216<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">LiveCodeBench v6</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">69.1</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">42.1</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">20.2</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">42.9</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">56.4</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">58.9</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">50.7</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">53.9</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">39.8</td></tr>217<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">LCB-Pro 25Q2 (Easy)</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">68.0</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">30.9</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">10.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">27.1</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">58.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">54.6</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">51.6</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">45.8</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">27.8</td></tr>218<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">LCB-Pro 25Q2 (Medium)</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">17.5</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">0.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">0.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">0.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">7.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">5.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">5.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">1.8</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">0.0</td></tr>219<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">OJBench</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">32.5</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">11.2</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">2.6</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">11.6</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">24.8</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">21.8</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">20.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">19.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">8.2</td></tr>220<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">SciCode (wbg)</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">26.3</strong><sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">14.2<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">2.8<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">20.9<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">16.1<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">24.9<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">16.4<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">24.4<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">7.8<sup style="font-size:0.72em;opacity:0.7">†</sup></td></tr>221<tr><td colspan="10" style="padding:5px 10px;font-weight:600;color:#1D6FD0;border-bottom:1px solid rgba(29, 111, 208, 0.2);background:rgba(29, 111, 208, 0.14)">Math Reasoning</td></tr>222<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">AIME 2025</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">86.5</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">41.9</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">29.6</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">31.7</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">78.8</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">79.4</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">56.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">37.1</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">46.0</td></tr>223<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">AIME 2026</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">86.5</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">45.2</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">29.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">39.8</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">82.7</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">83.5</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">62.1</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">45.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">56.7</td></tr>224<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">HMMT Feb 2026</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong>63.8</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">33.7</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">20.5</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">17.8</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">64.0</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">60.8</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">51.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">30.1</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">38.5</td></tr>225<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">MATH-500</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong>94.6</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">89.6</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">85.8</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">85.4</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">99.0</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">97.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">91.6</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">88.2</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">93.2</td></tr>226<tr><td colspan="10" style="padding:5px 10px;font-weight:600;color:#1D6FD0;border-bottom:1px solid rgba(29, 111, 208, 0.2);background:rgba(29, 111, 208, 0.14)">Instruction Following</td></tr>227<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">IFBench</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong>66.3</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">59.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">46.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">25.7</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">59.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">73.0</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">58.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">28.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">51.0</td></tr>228<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">IFEval</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;">86.7</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;"><strong>93.4</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">77.5</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">31.4</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">90.2</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">93.7</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">88.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">44.4</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">90.8</td></tr>229<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">Multi-IF</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;">71.8</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">76.8</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">57.1</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">40.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">73.6</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">75.9</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">65.9</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">45.9</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">71.4</td></tr>230<tr><td colspan="10" style="padding:5px 10px;font-weight:600;color:#1D6FD0;border-bottom:1px solid rgba(29, 111, 208, 0.2);background:rgba(29, 111, 208, 0.14)">General Knowledge</td></tr>231<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">MMLU-Pro</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong>70.8</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">65.2</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">64.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">56.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">78.0</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">65.8</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">65.7</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">68.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">63.1</td></tr>232<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">MMLU-Redux</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong>84.7</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">80.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">80.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">71.8</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">88.7</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">78.9</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">79.8</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">83.7</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">80.0</td></tr>233<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">HLE</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong>8.9</strong><sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">6.2<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">2.6<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">4.8<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">9.9</strong><sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">6.6<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">4.9<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">3.8<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">6.9<sup style="font-size:0.72em;opacity:0.7">†</sup></td></tr>234<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">GPQA-Diamond</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong>70.2</strong><sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">55.8<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">45.6<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">43.3<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">77.1</strong><sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">55.9<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">51.3<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">57.6<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">51.3<sup style="font-size:0.72em;opacity:0.7">†</sup></td></tr>235<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">SuperGPQA</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong>40.8</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">26.2</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">38.6</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">30.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">52.8</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">39.9</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">37.8</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">38.7</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">34.5</td></tr>236<tr><td colspan="10" style="padding:5px 10px;font-weight:600;color:#1D6FD0;border-bottom:1px solid rgba(29, 111, 208, 0.2);background:rgba(29, 111, 208, 0.14)">Long Context</td></tr>237<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">AA-LCR</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong>59.0</strong><sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">5.3<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">28.7<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">17.0<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">61.0</strong><sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">24.3<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">17.3<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">33.0<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">0.0<sup style="font-size:0.72em;opacity:0.7">†</sup></td></tr>238<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">NoLiMa</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">68.1</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">0.7</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">17.1</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">3.9</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">43.5</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">5.1</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">1.1</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">2.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">0.5</td></tr>239<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">LongBenchPro</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong>44.8</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">23.7</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">8.2</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">42.2</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">58.4</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">34.8</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">27.9</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">53.5</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">19.6</td></tr>240<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">LongBench v2</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong>43.7</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">30.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">24.9</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">33.2</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">47.3</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">36.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">32.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">42.7</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">30.4</td></tr>241<tr><td colspan="10" style="padding:5px 10px;font-weight:600;color:#1D6FD0;border-bottom:1px solid rgba(29, 111, 208, 0.2);background:rgba(29, 111, 208, 0.14)">Tool Use</td></tr>242<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">τ³-Bench Banking</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">20.8</strong><sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">7.2<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">2.1</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">3.9</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">6.8<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">5.6<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">1.2</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">4.1</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">3.4</td></tr>243<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">τ²-Bench Telecom</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">97.1</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">90.4</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">69.0<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">20.8<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">92.1<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">40.9</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">28.1<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">20.8<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">16.1<sup style="font-size:0.72em;opacity:0.7">†</sup></td></tr>244<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">BFCL v4</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">66.6</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">61.1</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">43.6</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">36.6</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">56.8</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">52.2</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">43.7</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">47.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">49.2</td></tr>245<tr><td colspan="10" style="padding:5px 10px;font-weight:600;color:#1D6FD0;border-bottom:1px solid rgba(29, 111, 208, 0.2);background:rgba(29, 111, 208, 0.14)">Coding Agent</td></tr>246<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">SWE-bench Verified</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">46.4</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">6.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">5.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">2.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">33.6</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">36.8</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">3.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">15.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">0.4</td></tr>247<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">SWE-bench Pro</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong>14.4</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">0.6</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">0.8</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">0.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">28.2</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">12.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">0.1</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">3.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">0.4</td></tr>248<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">Terminal-Bench v2.1</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong>8.6</strong><sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">4.5<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">3.0<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">0.4<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">25.8</strong><sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">13.9<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">3.8<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">1.9<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">1.9</td></tr>249<tr><td colspan="10" style="padding:5px 10px;font-weight:600;color:#1D6FD0;border-bottom:1px solid rgba(29, 111, 208, 0.2);background:rgba(29, 111, 208, 0.14)">Search Agent</td></tr>250<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">BrowseComp-ZH</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">43.5</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">9.8</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">18.2</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">4.7</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">39.6</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">21.1</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">3.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">7.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">13.2</td></tr>251<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">BrowseComp Top100</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">39.7</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">13.7</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">19.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">6.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">33.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">19.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">4.7</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">6.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">9.7</td></tr>252<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">GAIA Text-103</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">88.7</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">49.5</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">47.9</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">30.1</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">78.6</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">57.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">26.5</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">39.5</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">41.1</td></tr>253<tr><td colspan="10" style="padding:5px 10px;font-weight:600;color:#1D6FD0;border-bottom:1px solid rgba(29, 111, 208, 0.2);background:rgba(29, 111, 208, 0.14)">General Agent</td></tr>254<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">GDPval-AA v2</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">19.6</strong><sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">4.5</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">0.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">0.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">11.7</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">0.0<sup style="font-size:0.72em;opacity:0.7">†</sup></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">0.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">0.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">0.0</td></tr>255<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">Claw-Gym</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong>59.2</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">19.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">25.5</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">31.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">51.6</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">60.0</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">33.7</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">37.9</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">2.7</td></tr>256<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">WildClaw</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">23.9</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">10.2</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">9.2</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">8.9</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">17.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">20.0</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">8.9</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">14.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">4.5</td></tr>257<tr><td class="benchmark-cell" style="padding:5px 5px;padding-left:14px;border-bottom:1px solid rgba(128, 128, 128, 0.15);"><div class="benchmark-capability" style="font-size:14px;font-weight:600;line-height:1.22;color:#171717">QwenClaw</div></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);background:rgba(29, 111, 208, 0.08);vertical-align:middle;font-size:14px;line-height:1.2;"><strong style="color:#1D6FD0">42.9</strong></td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">19.3</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">18.2</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">14.5</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);border-left:1px solid rgba(29, 111, 208, 0.25);vertical-align:middle;font-size:14px;line-height:1.2;">37.1</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">36.4</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">16.8</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">16.7</td><td style="padding:5px 3px;text-align:center;border-bottom:1px solid rgba(128, 128, 128, 0.15);vertical-align:middle;font-size:14px;line-height:1.2;">4.5</td></tr>258</tbody></table>259 <p style="margin:6px 0 0;font-size:11px;line-height:1.55;opacity:0.75">1. <strong style="color:#1D6FD0">Blue bold</strong> indicates the best result across all models in the row (including 4B-class models); <strong>Black bold</strong> indicates the best result among 2B-class models.<br>2. Scores marked <sup style="font-size:0.72em;opacity:0.7">†</sup> come from the official Artificial Analysis release; all others are reproduced internally.</p>260</div>261 262## Training Recipe263 264The training of MiniCPM5-2B is a full-stack practice of **[UltraData Tiered Data Management](https://arxiv.org/pdf/2602.09003)**, covering three stages: base training, mid-training, and post-training.265 266During **base training**, the model goes through stable training and decay training to build core language capability and training stability. It then enters **mid-training** to further strengthen target capabilities and adapt to the target data distribution. The training corpus is released alongside the model as [Ultra-FineWeb](https://huggingface.co/datasets/openbmb/Ultra-FineWeb), [Ultra-FineWeb-L3](https://huggingface.co/datasets/openbmb/Ultra-FineWeb-L3), [UltraX](https://huggingface.co/datasets/openbmb/UltraX-Preview), [UltraData-Code](https://huggingface.co/datasets/openbmb/UltraData-Code) and [UltraData-Math](https://huggingface.co/datasets/openbmb/UltraData-Math).267 268During **post-training**, we proceed in three steps: **SFT**, **RL**, and **OPD**. We first use **400B tokens of deep-thinking SFT** to establish deep-thinking and general chat abilities; the SFT data is released as [UltraData-SFT-2605](https://huggingface.co/datasets/openbmb/UltraData-SFT-2605) and the Agent SFT data is released as [UltraData-SFT-Agent-2609](https://huggingface.co/datasets/openbmb/UltraData-SFT-Agent-2609). We then train specialized **RL teachers** for math, code, agentic tasks, writing, and related domains (with the corresponding data also open-sourced as [UltraData-RL-2609](https://huggingface.co/datasets/openbmb/UltraData-RL-2609)), and use **On-Policy Distillation (OPD)** to distill these teachers back into one release model.269 270271 272### What does RL + OPD bring?273 274**RL + OPD** is a key part of MiniCPM5-2B post-training. During the **RL** stage, we adopted the critic-based algorithm described in [JustRL II](https://panhaoxuan.notion.site/justrl-ii-scaling-small-llms-to-128k-reasoning-with-a-critic), substantially improving training stability and achieving significant gains across multiple domains. On the benchmarks listed below, RL + OPD improves reasoning and general capabilities by an average of **↑10.96 points**, and agentic capabilities by **↑6.96 points**.275 276**OPD** merges the capabilities of 16 expert models produced by RL training, including 5 agentic expert models. At each response position, we compute the full-vocabulary reverse KL divergence between student and teacher logits as the advantage estimate, replacing the original verification-based advantage. OPD directly reuses the prompts used to train each RL teacher as distillation data, so no additional corpus construction is required.277 278279 280## Quickstart281 282> [!Tip]283> We recommend using the following sets of sampling parameters for generation: `temperature=1.0, top_p=0.95, min_p=0.0`.284>285> If you encounter repetitive outputs, try: `temperature=1.0, top_p=0.95, min_p=0.0, repetition_penalty=1.05`.286>287> Please note that the support for sampling parameters varies according to inference frameworks.288 289### vLLM290 291```bash292pip install "vllm>=0.21"293vllm serve openbmb/MiniCPM5-2B --port 8000294```295 296```bash297curl http://localhost:8000/v1/chat/completions \298 -H "Content-Type: application/json" \299 -d '{300 "model": "openbmb/MiniCPM5-2B",301 "messages": [{"role": "user", "content": "Who are you? Please briefly introduce yourself."}],302 "max_tokens": 128,303 "temperature": 1.0304 }'305```306 307### SGLang308 309```bash310pip install "sglang[srt]>=0.5.16"311python -m sglang.launch_server --model-path openbmb/MiniCPM5-2B --port 30000312```313 314```bash315curl http://localhost:30000/v1/chat/completions \316 -H "Content-Type: application/json" \317 -d '{318 "model": "openbmb/MiniCPM5-2B",319 "messages": [{"role": "user", "content": "Who are you? Please briefly introduce yourself."}],320 "max_tokens": 128,321 "temperature": 1.0322 }'323```324 325**Speculative decoding (DSpark)**: we also release [MiniCPM5-2B-DSpark](https://huggingface.co/openbmb/MiniCPM5-2B-DSpark), a DSpark draft model trained for MiniCPM5-2B. Enable it in SGLang to accelerate decoding while keeping the target model's outputs unchanged:326 327```bash328python -m sglang.launch_server \329 --model-path openbmb/MiniCPM5-2B \330 --trust-remote-code \331 --speculative-algorithm DSPARK \332 --speculative-draft-model-path openbmb/MiniCPM5-2B-DSpark \333 --speculative-dspark-block-size 7 \334 --port 30000335```336 337### Llama.cpp338 339```bash340llama-server -m MiniCPM5-2B-F16.gguf -a MiniCPM5-2B --port 8080 -ngl 99 -c 8192 --jinja341```342 343`-c 8192` sets the context length. You can adjust this value as needed.344 345```bash346curl http://localhost:8080/v1/chat/completions \347 -H "Content-Type: application/json" \348 -d '{349 "model": "MiniCPM5-2B",350 "messages": [{"role": "user", "content": "1+1=?"}],351 "temperature": 1.0, "top_p": 0.95, "min_p": 0.0, "max_tokens": 256352 }'353```354 355In llama.cpp, the default `min_p=0.05` can lead to repetitive output: it filters out tokens whose probability is below 5% of the highest-probability token, potentially discarding the exact tokens needed to break out of a repetition loop. To prevent this, we set `min_p=0.0`.356 357### Transformers358 359```bash360pip install -U "transformers>=5.6" accelerate torch361```362 363```python364from transformers import AutoModelForCausalLM, AutoTokenizer365model_id = "openbmb/MiniCPM5-2B"366tokenizer = AutoTokenizer.from_pretrained(model_id)367model = AutoModelForCausalLM.from_pretrained(368 model_id,369 torch_dtype="auto",370 device_map="auto",371)372messages = [{"role": "user", "content": "Who are you? Please briefly introduce yourself."}]373inputs = tokenizer.apply_chat_template(374 messages,375 tokenize=True,376 add_generation_prompt=True,377 enable_thinking=True,378 return_dict=True,379 return_tensors="pt",380).to(model.device)381outputs = model.generate(**inputs, max_new_tokens=128)382print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:], skip_special_tokens=True))383```384 385## Tool Calling386 387For tool / function calling, **SGLang is the recommended backend**. MiniCPM5-2B emits XML-style tool calls and SGLang's built-in `minicpm5` parser converts them to OpenAI-compatible `tool_calls` natively:388 389```bash390python -m sglang.launch_server --model-path openbmb/MiniCPM5-2B --port 30000 \391 --tool-call-parser minicpm5 # or: --tool-call-parser auto392```393 394## GitHub Cookbooks and Agent Skills395 396MiniCPM5-2B uses the **standard `LlamaForCausalLM` architecture**, so mainstream inference engines can load it directly: **no custom kernels, no model-code fork**. For step-by-step deployment and fine-tuning instructions, use the GitHub cookbooks below. Agent Skills are linked as GitHub resources for users working with Cursor / Claude Code style coding agents.397 398### Deployment399 400| Backend | Model format / use case | Cookbook | Agent Skill |401| ------------ | ----------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------- |402| Transformers | BF16 / FP16 local Python inference, GPU + CPU | [transformers.md](https://github.com/OpenBMB/MiniCPM/blob/main/docs/deployment/transformers.md) | [minicpm5-deploy-transformers](https://github.com/OpenBMB/MiniCPM/blob/main/skills/minicpm5-deploy-transformers/SKILL.md) |403| vLLM | BF16 / FP16 OpenAI server | [vllm.md](https://github.com/OpenBMB/MiniCPM/blob/main/docs/deployment/vllm.md) | [minicpm5-deploy-vllm](https://github.com/OpenBMB/MiniCPM/blob/main/skills/minicpm5-deploy-vllm/SKILL.md) |404| SGLang | BF16 / FP16 OpenAI server, recommended for tool calling | [sglang.md](https://github.com/OpenBMB/MiniCPM/blob/main/docs/deployment/sglang.md) | [minicpm5-deploy-sglang](https://github.com/OpenBMB/MiniCPM/blob/main/skills/minicpm5-deploy-sglang/SKILL.md) |405| llama.cpp | GGUF local inference, CPU/GPU | [llama_cpp.md](https://github.com/OpenBMB/MiniCPM/blob/main/docs/deployment/llama_cpp.md) | [minicpm5-deploy-llama-cpp](https://github.com/OpenBMB/MiniCPM/blob/main/skills/minicpm5-deploy-llama-cpp/SKILL.md) |406| Ollama | GGUF local on-device runtime | [ollama.md](https://github.com/OpenBMB/MiniCPM/blob/main/docs/deployment/ollama.md) | [minicpm5-deploy-ollama](https://github.com/OpenBMB/MiniCPM/blob/main/skills/minicpm5-deploy-ollama/SKILL.md) |407| LM Studio | GGUF Mac desktop app and OpenAI server | [lmstudio.md](https://github.com/OpenBMB/MiniCPM/blob/main/docs/deployment/lmstudio.md) | [minicpm5-deploy-lmstudio](https://github.com/OpenBMB/MiniCPM/blob/main/skills/minicpm5-deploy-lmstudio/SKILL.md) |408| MLX | MLX / 4bit local inference on Apple Silicon | [mlx.md](https://github.com/OpenBMB/MiniCPM/blob/main/docs/deployment/mlx.md) | [minicpm5-deploy-mlx](https://github.com/OpenBMB/MiniCPM/blob/main/skills/minicpm5-deploy-mlx/SKILL.md) |409| ArcLight | GGUF local on-device, CPU, Desktop & Server | [arclight.md](https://github.com/OpenBMB/MiniCPM/blob/main/docs/deployment/arclight.md) | [minicpm5-deploy-arclight](https://github.com/OpenBMB/MiniCPM/blob/main/skills/minicpm5-deploy-arclight/SKILL.md) |410| vLLM Ascend | BF16 / FP16 OpenAI server | [vllm_ascend.md](https://github.com/OpenBMB/MiniCPM/blob/main/docs/deployment/vllm_ascend.md) | [minicpm5-deploy-vllm-ascend](https://github.com/OpenBMB/MiniCPM/blob/main/skills/minicpm5-deploy-vllm-ascend/SKILL.md) |411| LiteRT-LM | `.litertlm` on-device runtime: Android / iOS / desktop / IoT, CPU + GPU | [litert.md](https://github.com/OpenBMB/MiniCPM/blob/main/docs/deployment/litert.md) | [minicpm5-deploy-litert](https://github.com/OpenBMB/MiniCPM/blob/main/skills/minicpm5-deploy-litert/SKILL.md) |412 413### Fine-tuning414 415| Framework | Use case | Cookbook | Agent Skill |416| ------------- | ---------------------- | --------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------------------------------------- |417| TRL + PEFT | LoRA / SFT fine-tuning | [trl.md](https://github.com/OpenBMB/MiniCPM/blob/main/docs/finetune/trl.md) | [minicpm5-finetune-trl](https://github.com/OpenBMB/MiniCPM/blob/main/skills/minicpm5-finetune-trl/SKILL.md) |418| LLaMA-Factory | Fine-tuning | [llamafactory.md](https://github.com/OpenBMB/MiniCPM/blob/main/docs/finetune/llamafactory.md) | [minicpm5-finetune-llamafactory](https://github.com/OpenBMB/MiniCPM/blob/main/skills/minicpm5-finetune-llamafactory/SKILL.md) |419| ms-swift | Fine-tuning | [ms_swift.md](https://github.com/OpenBMB/MiniCPM/blob/main/docs/finetune/ms_swift.md) | [minicpm5-finetune-ms-swift](https://github.com/OpenBMB/MiniCPM/blob/main/skills/minicpm5-finetune-ms-swift/SKILL.md) |420| unsloth | Fine-tuning | [unsloth.md](https://github.com/OpenBMB/MiniCPM/blob/main/docs/finetune/unsloth.md) | [minicpm5-finetune-unsloth](https://github.com/OpenBMB/MiniCPM/blob/main/skills/minicpm5-finetune-unsloth/SKILL.md) |421 422### Other Supported Frameworks423 424In addition to the deployment and fine-tuning frameworks listed above, MiniCPM5-2B is also supported by FlagOS for multi-chip deployment.425 426#### FlagOS Overview427 428To enable large-scale deployment across different AI chips, Beijing Zhiyuan Research Institute, together with numerous research institutions, chip manufacturers, system vendors, and algorithm and software organizations both domestically and internationally, jointly initiated and established the FlagOS Open Source Community.429 430The FlagOS community is dedicated to building a unified, open-source system software stack for various AI chips, encompassing core open-source projects such as a large-scale operator library, a unified AI compiler, parallel training and inference frameworks, and a unified communication library. It aims to create an open technology ecosystem connecting the “model-system-chip” layers. By enabling “develop once, deploy across chips”, FlagOS unlocks the computational potential of hardware, breaks down the ecosystem silos between different chip software stacks, and effectively reduces migration costs for developers.The FlagOS community fosters an AI hardware and software ecosystem, overcomes single-vendor closed-source monopolies, promotes widespread deployment of AI hardware technologies, and is committed to rooted in China while embracing global collaboration.431 432Official website express: [https://flagos.io](https://flagos.io/)433 434<details>435<summary>FlagOS multi-chip support and usage</summary>436 437#### FlagOS: Supporting Multiple AI Chips438 439Thanks to FlagOS’s unified multi-chip AI system software stack, MiniCPM5-2B was adapted to 9 different AI chips in an extremely short time. Currently, the multi-chip version of MiniCPM5-2B has been released on FlagRelease, FlagOS’s platform for automatic migration, adaptation, and deployment of large models across multi-architecture AI chips. Details are as follows:440 441| Vendor | ModelScope | Huggingface |442| --------- | --------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------- |443| Nvidia | [MiniCPM5-2B-nvidia-FlagOS](https://www.modelscope.cn/models/FlagRelease/MiniCPM5-2B-nvidia-FlagOS) | [MiniCPM5-2B-nvidia-FlagOS](https://huggingface.co/FlagRelease/MiniCPM5-2B-nvidia-FlagOS) |444| Hygon | [MiniCPM5-2B-hygon-FlagOS](https://www.modelscope.cn/models/FlagRelease/MiniCPM5-2B-hygon-FlagOS) | [MiniCPM5-2B-hygon-FlagOS](https://huggingface.co/FlagRelease/MiniCPM5-2B-hygon-FlagOS) |445| Metax | [MiniCPM5-2B-metax-FlagOS](https://www.modelscope.cn/models/FlagRelease/MiniCPM5-2B-metax-FlagOS) | [MiniCPM5-2B-metax-FlagOS](https://huggingface.co/FlagRelease/MiniCPM5-2B-metax-FlagOS) |446| Iluvatar | [MiniCPM5-2B-iluvatar-FlagOS](https://www.modelscope.cn/models/FlagRelease/MiniCPM5-2B-iluvatar-FlagOS) | [MiniCPM5-2B-iluvatar-FlagOS](https://huggingface.co/FlagRelease/MiniCPM5-2B-iluvatar-FlagOS) |447| Zhenwu | [MiniCPM5-2B-zhenwu-FlagOS](https://www.modelscope.cn/models/FlagRelease/MiniCPM5-2B-zhenwu-FlagOS) | [MiniCPM5-2B-zhenwu-FlagOS](https://huggingface.co/FlagRelease/MiniCPM5-2B-zhenwu-FlagOS) |448| Mthreads | [MiniCPM5-2B-mthreads-FlagOS](https://www.modelscope.cn/models/FlagRelease/MiniCPM5-2B-mthreads-FlagOS) | [MiniCPM5-2B-mthreads-FlagOS](https://huggingface.co/FlagRelease/MiniCPM5-2B-mthreads-FlagOS) |449| Kunlunxin | [MiniCPM5-2B-kunlunxin-FlagOS](https://www.modelscope.cn/models/FlagRelease/MiniCPM5-2B-kunlunxin-FlagOS) | [MiniCPM5-2B-kunlunxin-FlagOS](https://huggingface.co/FlagRelease/MiniCPM5-2B-kunlunxin-FlagOS) |450| Ascend | [MiniCPM5-2B-ascend-FlagOS](https://modelscope.cn/models/FlagRelease/MiniCPM5-2B-ascend-FlagOS) | [MiniCPM5-2B-ascend-FlagOS](https://huggingface.co/FlagRelease/MiniCPM5-2B-ascend-FlagOS) |451| ARM-v9 | [MiniCPM5-2B-Armv9-FlagOS](https://modelscope.cn/models/FlagRelease/MiniCPM5-2B-Armv9-FlagOS) | [MiniCPM5-2B-Armv9-FlagOS](https://huggingface.co/FlagRelease/MiniCPM5-2B-Armv9-FlagOS) |452 453#### FlagOS Usage454 455##### FlagOS Performance Acceleration on Nvidia456 457###### From FlagRelease (**Recommendation**)458 459FlagRelease is a platform developed by the FlagOS team for automatic migration, adaptation, and deployment of large models across multi-architecture AI chips. The multi-chip version of MiniCPM5-2B has already been released on FlagRelease. All necessary software packages are pre-installed on the platform, so users do not need to install anything.460 461###### FlagRelease Image Key Versions462 463###### FlagRelease Quick Start464 465| Vendor | ModelScope | Huggingface |466| --------- | --------------------------------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------------- |467| Nvidia | [MiniCPM5-2B-nvidia-FlagOS](https://www.modelscope.cn/models/FlagRelease/MiniCPM5-2B-nvidia-FlagOS) | [MiniCPM5-2B-nvidia-FlagOS](https://huggingface.co/FlagRelease/MiniCPM5-2B-nvidia-FlagOS) |468| Hygon | [MiniCPM5-2B-hygon-FlagOS](https://www.modelscope.cn/models/FlagRelease/MiniCPM5-2B-hygon-FlagOS) | [MiniCPM5-2B-hygon-FlagOS](https://huggingface.co/FlagRelease/MiniCPM5-2B-hygon-FlagOS) |469| Metax | [MiniCPM5-2B-metax-FlagOS](https://www.modelscope.cn/models/FlagRelease/MiniCPM5-2B-metax-FlagOS) | [MiniCPM5-2B-metax-FlagOS](https://huggingface.co/FlagRelease/MiniCPM5-2B-metax-FlagOS) |470| Iluvatar | [MiniCPM5-2B-iluvatar-FlagOS](https://www.modelscope.cn/models/FlagRelease/MiniCPM5-2B-iluvatar-FlagOS) | [MiniCPM5-2B-iluvatar-FlagOS](https://huggingface.co/FlagRelease/MiniCPM5-2B-iluvatar-FlagOS) |471| Zhenwu | [MiniCPM5-2B-zhenwu-FlagOS](https://www.modelscope.cn/models/FlagRelease/MiniCPM5-2B-zhenwu-FlagOS) | [MiniCPM5-2B-zhenwu-FlagOS](https://huggingface.co/FlagRelease/MiniCPM5-2B-zhenwu-FlagOS) |472| Mthreads | [MiniCPM5-2B-mthreads-FlagOS](https://www.modelscope.cn/models/FlagRelease/MiniCPM5-2B-mthreads-FlagOS) | [MiniCPM5-2B-mthreads-FlagOS](https://huggingface.co/FlagRelease/MiniCPM5-2B-mthreads-FlagOS) |473| Kunlunxin | [MiniCPM5-2B-kunlunxin-FlagOS](https://www.modelscope.cn/models/FlagRelease/MiniCPM5-2B-kunlunxin-FlagOS) | [MiniCPM5-2B-kunlunxin-FlagOS](https://huggingface.co/FlagRelease/MiniCPM5-2B-kunlunxin-FlagOS) |474| Ascend | [MiniCPM5-2B-ascend-FlagOS](https://modelscope.cn/models/FlagRelease/MiniCPM5-2B-ascend-FlagOS) | [MiniCPM5-2B-ascend-FlagOS](https://huggingface.co/FlagRelease/MiniCPM5-2B-ascend-FlagOS) |475| ARM-v9 | [MiniCPM5-2B-Armv9-FlagOS](https://modelscope.cn/models/FlagRelease/MiniCPM5-2B-Armv9-FlagOS) | [MiniCPM5-2B-Armv9-FlagOS](https://huggingface.co/FlagRelease/MiniCPM5-2B-Armv9-FlagOS) |476 477###### From Scratch478 479- Dependencies: Python 3.12, GLIBC 2.39, GLIBCXX 3.4.33, CXXABI 1.3.15480 481###### Vllm Version482 483###### Installing the FlagOS Operator Library484 485Official Repository: https://github.com/flagos-ai/FlagGems486 487```PowerShell488pip install flag-gems==4.2.1rc0489pip install triton==3.5.1490```491 492###### Activating Acceleration493 494You can enable flagGems acceleration by adding the import of flagGems in the source code of vllm where inference is performed.495 496```Bash497import flag_gems498flag_gems.enable(record=True, once=True, path="/root/gems.txt")499```500 501```PowerShell502vllm serve ${model_path} \503--trust-remote-code \504--dtype bfloat16 \505--enforce-eager \506--port ${Port} \507--served-model-name ${model_name} \508--gpu-memory-utilization 0.85509```510 511##### Using FlagOS Unified Multi-Chip Backend Plugin512 513[**vllm-plugin-FL**](https://github.com/flagos-ai/vllm-plugin-FL) is a plugin built for the vLLM inference/service framework. Developed on top of FlagOS’s unified multi-chip backend, it is designed to extend vLLM’s capabilities and performance across a variety of hardware environments.514 515###### Using vllm-plugin-FL516 517| Vendor | From Scratch | From FlagRelease | |518| ------ | -------------------------------------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------ | ----------------------------------------------------------------------------------------- |519| Nvidia | [vllm-plugin-FL/MiniCPM5-2B](https://github.com/flagos-ai/vllm-plugin-FL/blob/main/examples/minicpm/README.md) | [MiniCPM5-2B-ModelScope](https://www.modelscope.cn/models/FlagRelease/MiniCPM5-2B-nvidia-FlagOS) | [MiniCPM5-2B-nvidia-FlagOS](https://huggingface.co/FlagRelease/MiniCPM5-2B-nvidia-FlagOS) |520 521</details>522 523## Limitations and Disclaimer524 525This model has no autonomous intent or legal personhood; its outputs are text generated from statistical patterns and may be inaccurate, biased, or offensive, and may be manipulated by carefully crafted prompts ("jailbreaks") into producing unintended content. Its responses on sensitive topics such as politics, health, finance, and law are not reviewed by experts and should not be treated as professional advice.526 527This model is provided "**AS IS**", without warranty of any kind, express or implied, and the developers are not liable for any damages arising from its use. Users must employ the model only for lawful, compliant, and ethical purposes, configure their own safeguards, and label AI-generated content where required; deliberate jailbreaking, injection attacks, or inducing harmful output is prohibited, and any such testing is at the user's own risk.528 529## License530 531This repository and MiniCPM model weights are released under the [Apache-2.0](https://github.com/OpenBMB/MiniCPM/blob/main/LICENSE) License.532 533## Citation534 535Please cite our paper if you find our work valuable:536 537```bibtex538@article{minicpm4,539 title={Minicpm4: Ultra-efficient llms on end devices},540 author={MiniCPM, Team},541 journal={arXiv preprint arXiv:2506.07900},542 year={2025}543}544```545