⚡ 速度 榜单排名
标准条件下的输出吞吐（tokens/s）。
🏆 综合673💻 编程392📐 数学32🧠 推理391🤖 智能体31📚 知识388👁️ 多模态370🛠️ 编码智能体40🔬 科学推理45🌍 多语言12✍️ 写作53💬 聊天助手53🔭 研究48🔧 工具调用31📜 长上下文55⚡ 速度378💰 性价比377🔓 开源权重153
共 378 个模型
| # | 模型名称 | Podium 得分 | 生成速度 |
|---|---|---|---|
| 01 | Gemini 3.5 Flash-Lite Google | 100.0 | 389 t/s |
| 02 | GPT-oss-120B OpenAI · 开源 | 84.0 | 240 t/s |
| 03 | Gemini 3.5 Flash Google | 65.5 | 267 t/s |
| 4 | Gemini 3.6 Flash Google | 55.9 | 233 t/s |
| 5 | Muse Spark 1.1 Meta | 48.9 | 208 t/s |
| 6 | Qwen3.7 Max Alibaba | 47.5 | 203 t/s |
| 7 | Claude Opus 5 Xhigh Anthropic | 40.0 | 80 t/s |
| 8 | Claude Opus 5 High Anthropic | 40.0 | 80 t/s |
| 9 | Grok 4 xAI | 40.0 | 80 t/s |
| 10 | Grok 4 6 xAI | 40.0 | 80 t/s |
| 11 | Claude Opus 5 Medium Anthropic | 40.0 | 80 t/s |
| 12 | Qwen3 8 2 4t A95b Alibaba · 开源 | 40.0 | 85 t/s |
| 13 | Gpt 5 6 Sol High OpenAI | 40.0 | 90 t/s |
| 14 | Muse Spark 1 2 Meta | 40.0 | 80 t/s |
| 15 | Gemini 3 7 Flash Google | 40.0 | 120 t/s |
| 16 | Gpt 5 6 Sol Medium OpenAI | 40.0 | 90 t/s |
| 17 | Gemini 3 7 Flash Medium Google | 40.0 | 120 t/s |
| 18 | Glm 5 Z.ai · 开源 | 40.0 | 60 t/s |
| 19 | Claude Opus 5 Low Anthropic | 40.0 | 80 t/s |
| 20 | Gpt 5 5 Medium OpenAI | 40.0 | 90 t/s |
| 21 | Gemini 3 7 Flash Low Google | 40.0 | 120 t/s |
| 22 | Gpt 5 6 Sol Low OpenAI | 40.0 | 90 t/s |
| 23 | Gpt 5 6 Terra High OpenAI | 40.0 | 90 t/s |
| 24 | Kimi K3 Low Moonshot AI | 40.0 | 60 t/s |
| 25 | Claude Sonnet 4 6 Adaptive Anthropic | 40.0 | 80 t/s |
| 26 | Gpt 5 6 Luna High OpenAI | 40.0 | 90 t/s |
| 27 | Gpt 5 6 Terra Medium OpenAI | 40.0 | 90 t/s |
| 28 | Deepseek V4 Pro 0424 DeepSeek · 开源 | 40.0 | 60 t/s |
| 29 | MiniMax-M2.7 MiniMax | 40.0 | 40 t/s |
| 30 | Claude Opus 4 6 Adaptive Anthropic | 40.0 | 80 t/s |
| 31 | Gpt 5 5 Low OpenAI | 40.0 | 90 t/s |
| 32 | Gpt 5 4 Mini OpenAI | 40.0 | 90 t/s |
| 33 | Claude Opus 4 7 Non Reasoning Anthropic | 40.0 | 50 t/s |
| 34 | Deepseek V4 Pro 0424 High DeepSeek · 开源 | 40.0 | 60 t/s |
| 35 | Claude Sonnet 5 Non Reasoning Anthropic | 40.0 | 50 t/s |
| 36 | Deepseek V4 Flash 0420 DeepSeek · 开源 | 40.0 | 60 t/s |
| 37 | Claude Opus 4 5 Thinking Anthropic | 40.0 | 50 t/s |
| 38 | 40.0 | 60 t/s | |
| 39 | Solar Pro4 Upstage | 40.0 | 40 t/s |
| 40 | Gpt 5 6 Terra Low OpenAI | 40.0 | 90 t/s |
| 41 | Gpt 5 2 Codex OpenAI | 40.0 | 90 t/s |
| 42 | Qwen3 6 Max Alibaba · 开源 | 40.0 | 85 t/s |
| 43 | 40.0 | 80 t/s | |
| 44 | Gpt 5 4 Low OpenAI | 40.0 | 90 t/s |
| 45 | Agnes 2 5 Pro Alpha Agnes | 40.0 | 60 t/s |
| 46 | Gpt 5 4 Nano OpenAI | 40.0 | 90 t/s |
| 47 | Deepseek V4 Flash 0420 High DeepSeek · 开源 | 40.0 | 60 t/s |
| 48 | Gpt 5 2 Medium OpenAI | 40.0 | 90 t/s |
| 49 | Gpt 5 6 Luna Medium OpenAI | 40.0 | 90 t/s |
| 50 | Gemini 3 Flash Reasoning Google | 40.0 | 80 t/s |
| 51 | 40.0 | 150 t/s | |
| 52 | Deepseek V3 2 Reasoning DeepSeek · 开源 | 40.0 | 40 t/s |
| 53 | Grok 4 20 xAI | 40.0 | 80 t/s |
| 54 | Mimo V2 5 0424 Xiaomi · 开源 | 40.0 | 60 t/s |
| 55 | Ling 3 0 Flash Ling | 40.0 | 60 t/s |
| 56 | Qwen3 6 27b Alibaba · 开源 | 40.0 | 85 t/s |
| 57 | Claude 4 5 Sonnet Thinking Anthropic | 40.0 | 50 t/s |
| 58 | 40.0 | 80 t/s | |
| 59 | Gpt 5 Codex OpenAI | 40.0 | 90 t/s |
| 60 | 40.0 | 80 t/s | |
| 61 | Glm 5 1 Non Reasoning Z.ai · 开源 | 40.0 | 40 t/s |
| 62 | Grok 4 3 Low xAI | 40.0 | 80 t/s |
| 63 | Kimi K2 5 Moonshot AI | 40.0 | 60 t/s |
| 64 | Gemini 3 5 Flash Minimal Google | 40.0 | 120 t/s |
| 65 | Gpt 5 5 Non Reasoning OpenAI | 40.0 | 60 t/s |
| 66 | Gpt 5 1 Codex OpenAI | 40.0 | 90 t/s |
| 67 | Kimi K2 6 Non Reasoning Moonshot AI | 40.0 | 40 t/s |
| 68 | 40.0 | 50 t/s | |
| 69 | Muse Glimmer Meta | 40.0 | 80 t/s |
| 70 | Mistral Medium 3.5 Mistral | 40.0 | 60 t/s |
| 71 | Glm 5 2 Non Reasoning Z.ai · 开源 | 40.0 | 40 t/s |
| 72 | 40.0 | 60 t/s | |
| 73 | Gpt 5 Medium OpenAI | 40.0 | 90 t/s |
| 74 | Claude 4 1 Opus Thinking Anthropic | 40.0 | 50 t/s |
| 75 | Hy3 Preview Tencent | 40.0 | 60 t/s |
| 76 | Gpt 5 5 Instant 05 26 OpenAI | 40.0 | 90 t/s |
| 77 | Gemini 3 Pro Low Google | 40.0 | 120 t/s |
| 78 | Gpt 5 6 Luna Low OpenAI | 40.0 | 90 t/s |
| 79 | Kimi K2 Thinking Moonshot AI | 40.0 | 40 t/s |
| 80 | O3 Pro OpenAI | 40.0 | 60 t/s |
| 81 | Glm 5 Non Reasoning Z.ai · 开源 | 40.0 | 40 t/s |
| 82 | Claude 4 5 Haiku Reasoning Anthropic | 40.0 | 50 t/s |
| 83 | Qwen3 5 397b A17b Non Reasoning Alibaba · 开源 | 40.0 | 55 t/s |
| 84 | Minimax M2 1 MiniMax · 开源 | 40.0 | 60 t/s |
| 85 | Qwen3 6 35b A3b Alibaba · 开源 | 40.0 | 85 t/s |
| 86 | 40.0 | 40 t/s | |
| 87 | 40.0 | 150 t/s | |
| 88 | Deepseek V4 Pro 0424 Non Reasoning DeepSeek · 开源 | 40.0 | 40 t/s |
| 89 | Gpt 5 Low OpenAI | 40.0 | 90 t/s |
| 90 | Mimo V2 Flash Reasoning Xiaomi · 开源 | 40.0 | 40 t/s |
| 91 | Claude 4 Opus Thinking Anthropic | 40.0 | 50 t/s |
| 92 | Gpt 5 Mini Medium OpenAI | 40.0 | 90 t/s |
| 93 | Gpt 5 1 Codex Mini OpenAI | 40.0 | 90 t/s |
| 94 | Qwen3 5 Omni Plus Alibaba · 开源 | 40.0 | 85 t/s |
| 95 | Qwen3 6 27b Non Reasoning Alibaba · 开源 | 40.0 | 55 t/s |
| 96 | Ring 2 6 1t WeRide | 40.0 | 60 t/s |
| 97 | Deepseek V3 1 Terminus Reasoning DeepSeek · 开源 | 40.0 | 40 t/s |
| 98 | Step 3 7 Flash StepFun | 40.0 | 60 t/s |
| 99 | Gpt 5 4 Nano Medium OpenAI | 40.0 | 90 t/s |
| 100 | Gpt 5 4 Mini Medium OpenAI | 40.0 | 90 t/s |
| 101 | Kimi K2 5 Non Reasoning Moonshot AI | 40.0 | 40 t/s |
| 102 | Qwen3 5 27b Non Reasoning Alibaba · 开源 | 40.0 | 55 t/s |
| 103 | Claude 4 5 Sonnet Anthropic | 40.0 | 80 t/s |
| 104 | Claude 4 Sonnet Thinking Anthropic | 40.0 | 50 t/s |
| 105 | Deepseek V4 Flash Non Reasoning DeepSeek · 开源 | 40.0 | 40 t/s |
| 106 | Glm 4 6 Reasoning Z.ai · 开源 | 40.0 | 40 t/s |
| 107 | Gpt 5 5 Instant 06 26 OpenAI | 40.0 | 90 t/s |
| 108 | Minimax M2 MiniMax · 开源 | 40.0 | 60 t/s |
| 109 | Claude 4 1 Opus Anthropic | 40.0 | 80 t/s |
| 110 | Mimo V2 5 Pro Non Reasoning Xiaomi · 开源 | 40.0 | 40 t/s |
| 111 | Gpt 5 4 Non Reasoning OpenAI | 40.0 | 60 t/s |
| 112 | Qwen3 5 122b A10b Non Reasoning Alibaba · 开源 | 40.0 | 55 t/s |
| 113 | 40.0 | 50 t/s | |
| 114 | Glm 4 7 Non Reasoning Z.ai · 开源 | 40.0 | 40 t/s |
| 115 | 40.0 | 60 t/s | |
| 116 | Hy3 Non Reasoning Tencent | 40.0 | 40 t/s |
| 117 | Ling 2 6 1t Ling | 40.0 | 60 t/s |
| 118 | Gpt 5 2 Non Reasoning OpenAI | 40.0 | 60 t/s |
| 119 | Gemma 4 26b A4b Google · 开源 | 40.0 | 120 t/s |
| 120 | O4 Mini OpenAI | 40.0 | 60 t/s |
| 121 | Claude 4 Opus Anthropic | 40.0 | 80 t/s |
| 122 | Claude 4 Sonnet Anthropic | 40.0 | 80 t/s |
| 123 | Step 3 5 Flash 0202 StepFun | 40.0 | 60 t/s |
| 124 | Deepseek V3 2 Reasoning 0925 DeepSeek · 开源 | 40.0 | 40 t/s |
| 125 | Gpt 5 Mini OpenAI | 40.0 | 90 t/s |
| 126 | Qwen3 Max Thinking Preview Alibaba · 开源 | 40.0 | 55 t/s |
| 127 | 40.0 | 50 t/s | |
| 128 | Qwen3 6 35b A3b Non Reasoning Alibaba · 开源 | 40.0 | 55 t/s |
| 129 | Ling 3 0 Tiny Ling | 40.0 | 60 t/s |
| 130 | Qwen3 5 35b A3b Non Reasoning Alibaba · 开源 | 40.0 | 55 t/s |
| 131 | Claude 4.5 Haiku Anthropic | 40.0 | 80 t/s |
| 132 | Kimi K2 0905 Moonshot AI | 40.0 | 60 t/s |
| 133 | Claude 3 7 Sonnet Anthropic | 40.0 | 80 t/s |
| 134 | O1 OpenAI | 40.0 | 60 t/s |
| 135 | Nemotron 3 5 Lightning NVIDIA | 40.0 | 150 t/s |
| 136 | Solar Pro 3 Upstage | 40.0 | 40 t/s |
| 137 | 40.0 | 50 t/s | |
| 138 | 40.0 | 50 t/s | |
| 139 | Command A Plus Cohere | 40.0 | 60 t/s |
| 140 | Gemini 2 5 Pro 05 06 Google | 40.0 | 120 t/s |
| 141 | Gemma 4 31b Non Reasoning Google · 开源 | 40.0 | 80 t/s |
| 142 | Gemma 4 12b Google · 开源 | 40.0 | 120 t/s |
| 143 | 40.0 | 50 t/s | |
| 144 | GPT-oss-20B OpenAI · 开源 | 40.0 | 90 t/s |
| 145 | 40.0 | 40 t/s | |
| 146 | Mercury 2 Inception | 40.0 | 60 t/s |
| 147 | Qwen3 5 9b Alibaba · 开源 | 40.0 | 85 t/s |
| 148 | Deepseek V3 2 0925 DeepSeek · 开源 | 40.0 | 60 t/s |
| 149 | Apriel V1 5 15b Thinker ServiceNow | 40.0 | 60 t/s |
| 150 | 40.0 | 40 t/s | |
| 151 | Qwen3 Coder Next Alibaba · 开源 | 40.0 | 85 t/s |
| 152 | Deepseek V3 1 Reasoning DeepSeek · 开源 | 40.0 | 40 t/s |
| 153 | Qwen3 Vl 235b A22b Reasoning Alibaba · 开源 | 40.0 | 55 t/s |
| 154 | Apriel V1 6 15b Thinker ServiceNow | 40.0 | 60 t/s |
| 155 | Nova 2 0 Lite Reasoning Amazon | 40.0 | 40 t/s |
| 156 | Gpt 5 1 Non Reasoning OpenAI | 40.0 | 60 t/s |
| 157 | Qwen3 5 9b Non Reasoning Alibaba · 开源 | 40.0 | 55 t/s |
| 158 | Gemma 4 26b A4b Non Reasoning Google · 开源 | 40.0 | 80 t/s |
| 159 | Qwen3 5 4b Alibaba · 开源 | 40.0 | 85 t/s |
| 160 | 40.0 | 80 t/s | |
| 161 | North Mini Code North | 40.0 | 60 t/s |
| 162 | Gpt 5 Nano OpenAI | 40.0 | 90 t/s |
| 163 | Qwen3 235b A22b Instruct 2507 Reasoning Alibaba · 开源 | 40.0 | 55 t/s |
| 164 | 40.0 | 40 t/s | |
| 165 | Kimi K2 Moonshot AI | 40.0 | 60 t/s |
| 166 | Mistral Small 4 Mistral · 开源 | 40.0 | 100 t/s |
| 167 | Gpt 4 1 OpenAI | 40.0 | 90 t/s |
| 168 | Qwen3 Max Preview Alibaba · 开源 | 40.0 | 85 t/s |
| 169 | Devstral 2 Mistral | 40.0 | 100 t/s |
| 170 | Gpt 5 Nano Medium OpenAI | 40.0 | 90 t/s |
| 171 | 40.0 | 40 t/s | |
| 172 | O3 Mini OpenAI | 40.0 | 60 t/s |
| 173 | Qwen3 5 Omni Flash Alibaba · 开源 | 40.0 | 85 t/s |
| 174 | O1 Pro OpenAI | 40.0 | 60 t/s |
| 175 | Deepseek R1 0120 DeepSeek · 开源 | 40.0 | 40 t/s |
| 176 | Grok 3 xAI | 40.0 | 80 t/s |
| 177 | Seed Oss 36b Instruct ByteDance | 40.0 | 60 t/s |
| 178 | Qwen3 235b A22b Instruct 2507 Alibaba · 开源 | 40.0 | 85 t/s |
| 179 | Trinity Large Thinking Trinity | 40.0 | 40 t/s |
| 180 | Qwen3 Coder 480b A35b Instruct Alibaba · 开源 | 40.0 | 85 t/s |
| 181 | Qwen3 Vl 32b Reasoning Alibaba · 开源 | 40.0 | 55 t/s |
| 182 | 40.0 | 40 t/s | |
| 183 | Minimax M1 80k MiniMax · 开源 | 40.0 | 60 t/s |
| 184 | 40.0 | 60 t/s | |
| 185 | Devstral Small 2 Mistral | 40.0 | 100 t/s |
| 186 | Gpt 5 Minimal OpenAI | 40.0 | 90 t/s |
| 187 | O1 Preview OpenAI | 40.0 | 60 t/s |
| 188 | Glm 4 6v Reasoning Z.ai · 开源 | 40.0 | 40 t/s |
| 189 | Qwen3 Next 80b A3b Reasoning Alibaba · 开源 | 40.0 | 55 t/s |
| 190 | 40.0 | 60 t/s | |
| 191 | 40.0 | 40 t/s | |
| 192 | Grok 4 Fast xAI | 40.0 | 80 t/s |
| 193 | Qwen3 5 4b Non Reasoning Alibaba · 开源 | 40.0 | 55 t/s |
| 194 | Mistral Large 3 Mistral · 开源 | 40.0 | 100 t/s |
| 195 | O3 Mini High OpenAI | 40.0 | 60 t/s |
| 196 | Glm 4 7 Flash Non Reasoning Z.ai · 开源 | 40.0 | 40 t/s |
| 197 | Deepseek V3 0324 DeepSeek · 开源 | 40.0 | 60 t/s |
| 198 | 40.0 | 80 t/s | |
| 199 | 40.0 | 150 t/s | |
| 200 | Gpt Oss 120b Low OpenAI | 40.0 | 90 t/s |
| 201 | Mistral Small 3 1 Mistral · 开源 | 40.0 | 100 t/s |
| 202 | Gpt 4 1 Mini OpenAI | 40.0 | 90 t/s |
| 203 | Mistral Medium 3 1 Mistral · 开源 | 40.0 | 100 t/s |
| 204 | Qwen3 30b A3b 2507 Reasoning Alibaba · 开源 | 40.0 | 55 t/s |
| 205 | Llama 4 Maverick Meta · 开源 | 40.0 | 80 t/s |
| 206 | 40.0 | 100 t/s | |
| 207 | Nova 2.0 Pro Preview Amazon · preview | 40.0 | 40 t/s |
| 208 | Gpt Oss 20b Low OpenAI | 40.0 | 90 t/s |
| 209 | Qwen3 Vl 235b A22b Instruct Alibaba · 开源 | 40.0 | 85 t/s |
| 210 | Gpt 5 Mini Minimal OpenAI | 40.0 | 90 t/s |
| 211 | Deepseek V3 DeepSeek · 开源 | 40.0 | 60 t/s |
| 212 | Ling 2 6 Flash Ling | 40.0 | 60 t/s |
| 213 | Qwen3 Next 80b A3b Instruct Alibaba · 开源 | 40.0 | 85 t/s |
| 214 | Qwen3 Coder 30b A3b Instruct Alibaba · 开源 | 40.0 | 85 t/s |
| 215 | Qwen3 235b A22b Instruct Reasoning Alibaba · 开源 | 40.0 | 55 t/s |
| 216 | Qwen3 Vl 30b A3b Reasoning Alibaba · 开源 | 40.0 | 55 t/s |
| 217 | Qwq 32b Alibaba | 40.0 | 85 t/s |
| 218 | Gemma 4 12b Non Reasoning Google · 开源 | 40.0 | 80 t/s |
| 219 | 40.0 | 120 t/s | |
| 220 | Nova Premier Amazon | 40.0 | 60 t/s |
| 221 | Mistral Medium 3 Mistral · 开源 | 40.0 | 100 t/s |
| 222 | Celeris 1 Celeris | 40.0 | 60 t/s |
| 223 | 40.0 | 50 t/s | |
| 224 | Mistral Small 4 Non Reasoning Mistral · 开源 | 40.0 | 60 t/s |
| 225 | Gemma 4 E4b Google · 开源 | 40.0 | 120 t/s |
| 226 | Claude 3 Opus Anthropic | 40.0 | 80 t/s |
| 227 | Nova 2 0 Lite Amazon | 40.0 | 60 t/s |
| 228 | 40.0 | 80 t/s | |
| 229 | Qwen3 32b Instruct Reasoning Alibaba · 开源 | 40.0 | 55 t/s |
| 230 | Ministral 3 14b Mistral | 40.0 | 100 t/s |
| 231 | Gpt 4o OpenAI | 40.0 | 90 t/s |
| 232 | Qwen3 Vl 32b Instruct Alibaba · 开源 | 40.0 | 85 t/s |
| 233 | Qwen3 235b A22b Instruct Alibaba · 开源 | 40.0 | 85 t/s |
| 234 | Mistral Small 3 2 Mistral · 开源 | 40.0 | 100 t/s |
| 235 | 40.0 | 120 t/s | |
| 236 | Qwen3 Vl 8b Reasoning Alibaba · 开源 | 40.0 | 55 t/s |
| 237 | Nova 2 0 Omni Amazon | 40.0 | 60 t/s |
| 238 | Qwen3 14b Instruct Reasoning Alibaba · 开源 | 40.0 | 55 t/s |
| 239 | Llama 4 Scout Meta · 开源 | 40.0 | 80 t/s |
| 240 | Hermes 4 Llama 3 1 70b Reasoning Meta · 开源 | 40.0 | 50 t/s |
| 241 | Qwen3 Vl 30b A3b Instruct Alibaba · 开源 | 40.0 | 85 t/s |
| 242 | Claude 35 Sonnet Anthropic | 40.0 | 80 t/s |
| 243 | Deepseek R1 Distill Llama 70b DeepSeek · 开源 | 40.0 | 40 t/s |
| 244 | Gpt 4 1 Nano OpenAI | 40.0 | 90 t/s |
| 245 | Qwen3 Omni 30b A3b Reasoning Alibaba · 开源 | 40.0 | 55 t/s |
| 246 | Gpt 4o 2024 08 06 OpenAI | 40.0 | 90 t/s |
| 247 | Llama 3 3 Instruct 70b Meta · 开源 | 40.0 | 80 t/s |
| 248 | Qwen3 30b A3b Instruct Reasoning Alibaba · 开源 | 40.0 | 55 t/s |
| 249 | Glm 4 5v Reasoning Z.ai · 开源 | 40.0 | 40 t/s |
| 250 | Ministral 3 8b Mistral | 40.0 | 100 t/s |
| 251 | Ernie 4 5 300b A47b Baidu | 40.0 | 60 t/s |
| 252 | 40.0 | 50 t/s | |
| 253 | Qwen3 30b A3b 2507 Alibaba · 开源 | 40.0 | 85 t/s |
| 254 | Hermes 4 Llama 3 1 405b Reasoning Meta · 开源 | 40.0 | 50 t/s |
| 255 | 40.0 | 100 t/s | |
| 256 | Gemma 4 E4b Non Reasoning Google · 开源 | 40.0 | 80 t/s |
| 257 | 40.0 | 100 t/s | |
| 258 | Hermes 4 Llama 3 1 405b Meta · 开源 | 40.0 | 80 t/s |
| 259 | Llama Nemotron Super 49b V1 5 Meta · 开源 | 40.0 | 80 t/s |
| 260 | Qwen3 32b Instruct Alibaba · 开源 | 40.0 | 85 t/s |
| 261 | Gpt 4o 2024 05 13 OpenAI | 40.0 | 90 t/s |
| 262 | Qwen3 8b Instruct Reasoning Alibaba · 开源 | 40.0 | 55 t/s |
| 263 | Qwen3 Vl 8b Instruct Alibaba · 开源 | 40.0 | 85 t/s |
| 264 | Claude 35 Sonnet June 24 Anthropic | 40.0 | 80 t/s |
| 265 | Gpt 5 Nano Minimal OpenAI | 40.0 | 90 t/s |
| 266 | Gpt 4 Turbo OpenAI | 40.0 | 90 t/s |
| 267 | Command A Cohere | 40.0 | 60 t/s |
| 268 | Nova Pro Amazon | 40.0 | 60 t/s |
| 269 | Gemma 3 27b Google · 开源 | 40.0 | 120 t/s |
| 270 | Llama 3 1 Instruct 8b Meta · 开源 | 40.0 | 80 t/s |
| 271 | Llama 3 1 Nemotron Instruct 70b Meta · 开源 | 40.0 | 80 t/s |
| 272 | 40.0 | 150 t/s | |
| 273 | 40.0 | 150 t/s | |
| 274 | Ministral 3 3b Mistral | 40.0 | 100 t/s |
| 275 | Mistral Large 2407 Mistral · 开源 | 40.0 | 100 t/s |
| 276 | Gpt 4 OpenAI | 40.0 | 90 t/s |
| 277 | Qwen3 14b Instruct Alibaba · 开源 | 40.0 | 85 t/s |
| 278 | Gemini 2 5 Flash Lite Google | 40.0 | 120 t/s |
| 279 | Gpt 4o Mini OpenAI | 40.0 | 90 t/s |
| 280 | Hermes 4 Llama 3 1 70b Meta · 开源 | 40.0 | 80 t/s |
| 281 | Mistral Small 3 Mistral · 开源 | 40.0 | 100 t/s |
| 282 | Nova Lite Amazon | 40.0 | 60 t/s |
| 283 | Qwen3 30b A3b Instruct Alibaba · 开源 | 40.0 | 85 t/s |
| 284 | Llama 3 1 Instruct 70b Meta · 开源 | 40.0 | 80 t/s |
| 285 | Qwen Turbo Alibaba · 开源 | 40.0 | 85 t/s |
| 286 | Reka Flash Reka | 40.0 | 60 t/s |
| 287 | Solar Mini Upstage | 40.0 | 60 t/s |
| 288 | Phi 4 Mini Microsoft · 开源 | 40.0 | 60 t/s |
| 289 | Gemma 3 12b Google · 开源 | 40.0 | 120 t/s |
| 290 | Granite 4 0 H Small IBM · 开源 | 40.0 | 60 t/s |
| 291 | Hermes 3 Llama 3 1 70b Meta · 开源 | 40.0 | 80 t/s |
| 292 | Jamba 1 5 Large AI21 Labs | 40.0 | 60 t/s |
| 293 | Qwen3 8b Instruct Alibaba · 开源 | 40.0 | 85 t/s |
| 294 | Qwen3 Omni 30b A3b Instruct Alibaba · 开源 | 40.0 | 85 t/s |
| 295 | Phi 4 Microsoft · 开源 | 40.0 | 60 t/s |
| 296 | Nova Micro Amazon | 40.0 | 60 t/s |
| 297 | Mistral Small Mistral · 开源 | 40.0 | 100 t/s |
| 298 | 40.0 | 150 t/s | |
| 299 | Mistral Large Mistral · 开源 | 40.0 | 100 t/s |
| 300 | Llama 2 Chat 7b Meta · 开源 | 40.0 | 80 t/s |
| 301 | Reka Flash 3 Reka | 40.0 | 60 t/s |
| 302 | Claude 3 Haiku Anthropic | 40.0 | 80 t/s |
| 303 | Gpt 35 Turbo OpenAI | 40.0 | 90 t/s |
| 304 | Mistral Medium Mistral · 开源 | 40.0 | 100 t/s |
| 305 | Mistral Small 2402 Mistral · 开源 | 40.0 | 100 t/s |
| 306 | Llama 3 Instruct 70b Meta · 开源 | 40.0 | 80 t/s |
| 307 | Llama 3 2 Instruct 11b Vision Meta · 开源 | 40.0 | 80 t/s |
| 308 | Command R Plus 04 2024 Cohere | 40.0 | 60 t/s |
| 309 | Olmo 3 7b Instruct AI2 · 开源 | 40.0 | 60 t/s |
| 310 | Jamba 1 5 Mini AI21 Labs | 40.0 | 60 t/s |
| 311 | Apertus 70b Instruct Swiss AI | 40.0 | 60 t/s |
| 312 | Mixtral 8x7b Instruct Mistral | 40.0 | 100 t/s |
| 313 | Command R 03 2024 Cohere | 40.0 | 60 t/s |
| 314 | Mistral 7b Instruct Mistral · 开源 | 40.0 | 100 t/s |
| 315 | Granite 3 3 8b Instruct IBM · 开源 | 40.0 | 60 t/s |
| 316 | Apertus 8b Instruct Swiss AI | 40.0 | 60 t/s |
| 317 | Gemma 3 1b Google · 开源 | 40.0 | 120 t/s |
| 318 | Gemma 3 4b Google · 开源 | 40.0 | 120 t/s |
| 319 | Gemma 3n E2b Google · 开源 | 40.0 | 120 t/s |
| 320 | Gemma 3n E4b Google · 开源 | 40.0 | 120 t/s |
| 321 | Llama 3 Instruct 8b Meta · 开源 | 40.0 | 80 t/s |
| 322 | Tiny Aya Global Cohere | 40.0 | 60 t/s |
| 323 | Claude Mythos 5 Anthropic · preview | 40.0 | 50 t/s |
| 324 | GPT-5.5 Pro OpenAI | 40.0 | 60 t/s |
| 325 | Claude Sonnet 4.6 (max) Anthropic | 40.0 | 50 t/s |
| 326 | Amazon Nova Amazon | 40.0 | 60 t/s |
| 327 | Async Flash V1 5 Async | 40.0 | 60 t/s |
| 328 | Async Pro V1 0 Async | 40.0 | 60 t/s |
| 329 | Asyncflow V2 Async | 40.0 | 60 t/s |
| 330 | Claude Sonnet 5 High Anthropic | 40.0 | 80 t/s |
| 331 | Claude Sonnet 5 Low Anthropic | 40.0 | 80 t/s |
| 332 | Claude Sonnet 5 Medium Anthropic | 40.0 | 80 t/s |
| 333 | Claude Sonnet 5 Xhigh Anthropic | 40.0 | 80 t/s |
| 334 | Cogito V2 1 Reasoning Cogito | 40.0 | 40 t/s |
| 335 | Falcon 2 TII · 开源 | 40.0 | 60 t/s |
| 336 | Gemini Google | 40.0 | 120 t/s |
| 337 | 40.0 | 120 t/s | |
| 338 | Gemini_nano Banana Pro Google | 40.0 | 120 t/s |
| 339 | Gladia Solaria Upstage | 40.0 | 60 t/s |
| 340 | Gpt 5 4 Pro OpenAI | 40.0 | 90 t/s |
| 341 | Gpt Realtime 2 OpenAI | 40.0 | 90 t/s |
| 342 | Grok Stt xAI | 40.0 | 80 t/s |
| 343 | Hidream O1 Image OpenAI | 40.0 | 60 t/s |
| 344 | Minimax H3 MiniMax · 开源 | 40.0 | 60 t/s |
| 345 | Mistral Mistral · 开源 | 40.0 | 100 t/s |
| 346 | Nvidia NVIDIA | 40.0 | 150 t/s |
| 347 | 40.0 | 150 t/s | |
| 348 | Ocateve 2 Ocat | 40.0 | 60 t/s |
| 349 | Qwen Image Alibaba · 开源 | 40.0 | 85 t/s |
| 350 | Qwen3 Asr Alibaba · 开源 | 40.0 | 85 t/s |
| 351 | GPT-5.6 Luna OpenAI | 39.5 | 175 t/s |
| 352 | GLM-5.2 Z.ai · 开源 | 36.4 | 164 t/s |
| 353 | Nemotron 3 Ultra 550B NVIDIA · 开源 | 30.2 | 142 t/s |
| 354 | Nex N2 Pro Nex AGI | 29.1 | 138 t/s |
| 355 | Gemini 3.1 Pro Google | 28.5 | 136 t/s |
| 356 | Inkling Small Thinking Machines | 26.8 | 130 t/s |
| 357 | Grok 4.3 xAI | 26.6 | 129 t/s |
| 358 | GPT-5.6 Terra OpenAI | 26.0 | 127 t/s |
| 359 | GLM-5.1 Z.ai · 开源 | 26.0 | 74 t/s |
| 360 | Qwen3.5 122B-A10B Alibaba · 开源 | 25.7 | 126 t/s |
| 361 | GPT-5.3 Codex OpenAI | 23.4 | 118 t/s |
| 362 | DeepSeek V4 Flash DeepSeek · 开源 | 22.0 | 113 t/s |
| 363 | Claude Sonnet 5 Anthropic | 17.2 | 96 t/s |
| 364 | MiniMax M3 MiniMax | 16.4 | 93 t/s |
| 365 | Inkling Thinking Machines | 13.0 | 81 t/s |
| 366 | GPT-5.6 Sol OpenAI | 10.5 | 72 t/s |
| 367 | Claude Fable 5 Anthropic | 10.2 | 71 t/s |
| 368 | Qwen3.5 397B-A17B Alibaba · 开源 | 10.2 | 71 t/s |
| 369 | Hunyuan Hy3 Tencent | 9.6 | 69 t/s |
| 370 | DeepSeek V4 Pro DeepSeek · 开源 | 8.8 | 66 t/s |
| 371 | MiMo V2.5 Pro Xiaomi · 开源 | 8.5 | 65 t/s |
| 372 | Claude Opus 5 Anthropic | 7.1 | 60 t/s |
| 373 | Grok 4.5 xAI | 6.8 | 59 t/s |
| 374 | Qwen3.6 Plus Alibaba | 5.6 | 55 t/s |
| 375 | Qwen3.7 Plus Alibaba | 5.4 | 54 t/s |
| 376 | Kimi K2.7 Code Moonshot AI · 开源 | 1.1 | 39 t/s |
| 377 | Kimi K3 Moonshot AI · 开源 | 0.6 | 37 t/s |
| 378 | Gemma 4 31B Google · 开源 | 0.0 | 35 t/s |
Podium Score是整合5个榜单的加权综合分。详见评分方法论。