{ "rankings": [ { "id": "llm-week", "title": "LLM 周榜", "description": "查看最近一周的模型热度与用量变化。", "category": "真实使用", "sourceLabel": "公开榜单页面与模型目录", "sourceUrl": "https://openrouter.ai/rankings", "updatedAt": "2026-03-26T07:00:14.161Z", "methodology": "直接抓取公开榜单页面中的 Top Models 区块，并补充模型上下文和说明。", "items": [ { "rank": 1, "model": "Mimo V2 Pro", "provider": "Xiaomi", "score": "2.97T", "change": null, "summary": "MiMo-V2-Pro is Xiaomi's flagship foundation model, featuring over 1T total parameters and a 1M context length, deeply optimized for agent…", "link": "https://openrouter.ai/xiaomi/mimo-v2-pro-20260318", "metrics": { "window": "7d", "tokens": "2.97T", "change": "new", "context": "1.0M" }, "secondary_en": "Xiaomi", "secondary_zh": "Xiaomi", "summary_zh": "MiMo-V2-Pro是小米的旗舰基础模型，拥有超过1万亿总参数和100万上下文长度，专为智能体场景深度优化……", "summary_en": "MiMo-V2-Pro is Xiaomi's flagship foundation model, featuring over 1T total parameters and a 1M context length, deeply optimized for agent…" }, { "rank": 2, "model": "Step 3.5 Flash (free)", "provider": "StepFun", "score": "1.57T", "change": 17, "summary": "Step 3.5 Flash is StepFun's most capable open-source foundation model. Built on a sparse Mixture of Experts (MoE) architecture, it select…", "link": "https://openrouter.ai/stepfun/step-3.5-flash:free", "metrics": { "window": "7d", "tokens": "1.57T", "change": "+17%", "context": "256K" }, "secondary_en": "StepFun", "secondary_zh": "StepFun", "summary_zh": "Step 3.5 Flash是StepFun最强大的开源基础模型。它基于稀疏专家混合（MoE）架构构建，能够智能选择……", "summary_en": "Step 3.5 Flash is StepFun's most capable open-source foundation model. Built on a sparse Mixture of Experts (MoE) architecture, it select…" }, { "rank": 3, "model": "Deepseek V3.2", "provider": "DeepSeek", "score": "1.18T", "change": 8, "summary": "DeepSeek-V3.2 is a large language model designed to harmonize high computational efficiency with strong reasoning and agentic tool-use pe…", "link": "https://openrouter.ai/deepseek/deepseek-v3.2-20251201", "metrics": { "window": "7d", "tokens": "1.18T", "change": "+8%", "context": "164K" }, "secondary_en": "DeepSeek", "secondary_zh": "DeepSeek", "summary_zh": "DeepSeek-V3.2是一款大型语言模型，旨在实现高计算效率与强大推理及智能体工具使用能力的完美结合……", "summary_en": "DeepSeek-V3.2 is a large language model designed to harmonize high computational efficiency with strong reasoning and agentic tool-use pe…" }, { "rank": 4, "model": "Minimax M2.5", "provider": "MiniMax", "score": "1.1T", "change": -28, "summary": "MiniMax-M2.5 is a SOTA large language model designed for real-world productivity. Trained in a diverse range of complex real-world digita…", "link": "https://openrouter.ai/minimax/minimax-m2.5-20260211", "metrics": { "window": "7d", "tokens": "1.1T", "change": "-28%", "context": "197K" }, "secondary_en": "MiniMax", "secondary_zh": "MiniMax", "summary_zh": "MiniMax-M2.5是一款专为现实世界生产力设计的先进大型语言模型。它通过多样化的复杂现实世界数字场景训练而成…", "summary_en": "MiniMax-M2.5 is a SOTA large language model designed for real-world productivity. Trained in a diverse range of complex real-world digita…" }, { "rank": 5, "model": "GLM 5 Turbo", "provider": "Z.ai", "score": "1.08T", "change": 189, "summary": "GLM-5 Turbo is a new model from Z.ai designed for fast inference and strong performance in agent-driven environments such as OpenClaw sce…", "link": "https://openrouter.ai/z-ai/glm-5-turbo-20260315", "metrics": { "window": "7d", "tokens": "1.08T", "change": "+189%", "context": "203K" }, "secondary_en": "Z.ai", "secondary_zh": "Z.ai", "summary_zh": "GLM-5 Turbo是Z.ai推出的新模型，专为快速推理和在OpenClaw等智能体驱动环境中实现强劲性能而设计…", "summary_en": "GLM-5 Turbo is a new model from Z.ai designed for fast inference and strong performance in agent-driven environments such as OpenClaw sce…" }, { "rank": 6, "model": "Claude Sonnet 4.6", "provider": "Anthropic", "score": "1.04T", "change": 7, "summary": "Sonnet 4.6 is Anthropic's most capable Sonnet-class model yet, with frontier performance across coding, agents, and professional work. It…", "link": "https://openrouter.ai/anthropic/claude-4.6-sonnet-20260217", "metrics": { "window": "7d", "tokens": "1.04T", "change": "+7%", "context": "1.0M" }, "secondary_en": "Anthropic", "secondary_zh": "Anthropic", "summary_zh": "Sonnet 4.6是Anthropic迄今为止最强大的Sonnet系列模型，在编程、智能体应用和专业工作领域均展现出前沿性能。它……", "summary_en": "Sonnet 4.6 is Anthropic's most capable Sonnet-class model yet, with frontier performance across coding, agents, and professional work. It…" }, { "rank": 7, "model": "Claude Opus 4.6", "provider": "Anthropic", "score": "1.02T", "change": 11, "summary": "Opus 4.6 is Anthropic’s strongest model for coding and long-running professional tasks. It is built for agents that operate across entire…", "link": "https://openrouter.ai/anthropic/claude-4.6-opus-20260205", "metrics": { "window": "7d", "tokens": "1.02T", "change": "+11%", "context": "1.0M" }, "secondary_en": "Anthropic", "secondary_zh": "Anthropic", "summary_zh": "Opus 4.6是Anthropic针对编程和长时间专业任务推出的最强模型。它专为能在整个...范围内运行的智能体而构建。", "summary_en": "Opus 4.6 is Anthropic’s strongest model for coding and long-running professional tasks. It is built for agents that operate across entire…" }, { "rank": 8, "model": "Gemini 3 Flash Preview", "provider": "Google", "score": "944B", "change": -8, "summary": "Gemini 3 Flash Preview is a high speed, high value thinking model designed for agentic workflows, multi turn chat, and coding assistance.…", "link": "https://openrouter.ai/google/gemini-3-flash-preview-20251217", "metrics": { "window": "7d", "tokens": "944B", "change": "-8%", "context": "1.0M" }, "secondary_en": "Google", "secondary_zh": "Google", "summary_zh": "Gemini 3 Flash Preview是一款专为智能体工作流、多轮对话及编程辅助设计的高速、高性价比思维模型。", "summary_en": "Gemini 3 Flash Preview is a high speed, high value thinking model designed for agentic workflows, multi turn chat, and coding assistance.…" }, { "rank": 9, "model": "Minimax M2.7", "provider": "MiniMax", "score": "926B", "change": null, "summary": "MiniMax-M2.7 is a next-generation large language model designed for autonomous, real-world productivity and continuous improvement. Built…", "link": "https://openrouter.ai/minimax/minimax-m2.7-20260318", "metrics": { "window": "7d", "tokens": "926B", "change": "-", "context": "205K" }, "secondary_en": "MiniMax", "secondary_zh": "MiniMax", "summary_zh": "MiniMax-M2.7是新一代大型语言模型，专为自主、现实世界生产力及持续改进而设计。构建……", "summary_en": "MiniMax-M2.7 is a next-generation large language model designed for autonomous, real-world productivity and continuous improvement. Built…" }, { "rank": 10, "model": "Kimi K2.5 0127", "provider": "MoonshotAI", "score": "558B", "change": -3, "summary": "Kimi K2.5 is Moonshot AI's native multimodal model, delivering state-of-the-art visual coding capability and a self-directed agent swarm…", "link": "https://openrouter.ai/moonshotai/kimi-k2.5-0127", "metrics": { "window": "7d", "tokens": "558B", "change": "-3%", "context": "262K" }, "secondary_en": "MoonshotAI", "secondary_zh": "MoonshotAI", "summary_zh": "Kimi K2.5是月之暗面AI自主研发的原生多模态模型，具备顶尖的视觉编码能力与自主智能体集群…", "summary_en": "Kimi K2.5 is Moonshot AI's native multimodal model, delivering state-of-the-art visual coding capability and a self-directed agent swarm…" } ] }, { "id": "llm-day", "title": "LLM 日榜", "description": "查看最近一天的模型热度与用量变化。", "category": "真实使用", "sourceLabel": "公开榜单页面与模型目录", "sourceUrl": "https://openrouter.ai/rankings?view=day", "updatedAt": "2026-03-26T07:00:14.162Z", "methodology": "直接抓取公开榜单页面中的 Top Models 区块，并补充模型上下文和说明。", "items": [ { "rank": 1, "model": "Mimo V2 Pro", "provider": "Xiaomi", "score": "481B", "change": -5, "summary": "MiMo-V2-Pro is Xiaomi's flagship foundation model, featuring over 1T total parameters and a 1M context length, deeply optimized for agent…", "link": "https://openrouter.ai/xiaomi/mimo-v2-pro-20260318", "metrics": { "window": "24h", "tokens": "481B", "change": "-5%", "context": "1.0M" }, "secondary_en": "Xiaomi", "secondary_zh": "Xiaomi", "summary_zh": "MiMo-V2-Pro是小米的旗舰基础模型，拥有超过1万亿总参数和100万上下文长度，专为智能体场景深度优化……", "summary_en": "MiMo-V2-Pro is Xiaomi's flagship foundation model, featuring over 1T total parameters and a 1M context length, deeply optimized for agent…" }, { "rank": 2, "model": "Step 3.5 Flash (free)", "provider": "StepFun", "score": "225B", "change": -1, "summary": "Step 3.5 Flash is StepFun's most capable open-source foundation model. Built on a sparse Mixture of Experts (MoE) architecture, it select…", "link": "https://openrouter.ai/stepfun/step-3.5-flash:free", "metrics": { "window": "24h", "tokens": "225B", "change": "-1%", "context": "256K" }, "secondary_en": "StepFun", "secondary_zh": "StepFun", "summary_zh": "Step 3.5 Flash是StepFun最强大的开源基础模型。它基于稀疏专家混合（MoE）架构构建，能够智能选择……", "summary_en": "Step 3.5 Flash is StepFun's most capable open-source foundation model. Built on a sparse Mixture of Experts (MoE) architecture, it select…" }, { "rank": 3, "model": "Minimax M2.7", "provider": "MiniMax", "score": "211B", "change": 7, "summary": "MiniMax-M2.7 is a next-generation large language model designed for autonomous, real-world productivity and continuous improvement. Built…", "link": "https://openrouter.ai/minimax/minimax-m2.7-20260318", "metrics": { "window": "24h", "tokens": "211B", "change": "+7%", "context": "205K" }, "secondary_en": "MiniMax", "secondary_zh": "MiniMax", "summary_zh": "MiniMax-M2.7是新一代大型语言模型，专为自主、现实世界生产力及持续改进而设计。构建……", "summary_en": "MiniMax-M2.7 is a next-generation large language model designed for autonomous, real-world productivity and continuous improvement. Built…" }, { "rank": 4, "model": "Claude Sonnet 4.6", "provider": "Anthropic", "score": "177B", "change": 5, "summary": "Sonnet 4.6 is Anthropic's most capable Sonnet-class model yet, with frontier performance across coding, agents, and professional work. It…", "link": "https://openrouter.ai/anthropic/claude-4.6-sonnet-20260217", "metrics": { "window": "24h", "tokens": "177B", "change": "+5%", "context": "1.0M" }, "secondary_en": "Anthropic", "secondary_zh": "Anthropic", "summary_zh": "Sonnet 4.6是Anthropic迄今为止最强大的Sonnet系列模型，在编程、智能体应用和专业工作领域均展现出前沿性能。它……", "summary_en": "Sonnet 4.6 is Anthropic's most capable Sonnet-class model yet, with frontier performance across coding, agents, and professional work. It…" }, { "rank": 5, "model": "Deepseek V3.2", "provider": "DeepSeek", "score": "177B", "change": -1, "summary": "DeepSeek-V3.2 is a large language model designed to harmonize high computational efficiency with strong reasoning and agentic tool-use pe…", "link": "https://openrouter.ai/deepseek/deepseek-v3.2-20251201", "metrics": { "window": "24h", "tokens": "177B", "change": "-1%", "context": "164K" }, "secondary_en": "DeepSeek", "secondary_zh": "DeepSeek", "summary_zh": "DeepSeek-V3.2是一款大型语言模型，旨在实现高计算效率与强大推理及智能体工具使用能力的完美结合……", "summary_en": "DeepSeek-V3.2 is a large language model designed to harmonize high computational efficiency with strong reasoning and agentic tool-use pe…" }, { "rank": 6, "model": "Claude Opus 4.6", "provider": "Anthropic", "score": "164B", "change": -2, "summary": "Opus 4.6 is Anthropic’s strongest model for coding and long-running professional tasks. It is built for agents that operate across entire…", "link": "https://openrouter.ai/anthropic/claude-4.6-opus-20260205", "metrics": { "window": "24h", "tokens": "164B", "change": "-2%", "context": "1.0M" }, "secondary_en": "Anthropic", "secondary_zh": "Anthropic", "summary_zh": "Opus 4.6是Anthropic针对编程和长时间专业任务推出的最强模型。它专为能在整个...范围内运行的智能体而构建。", "summary_en": "Opus 4.6 is Anthropic’s strongest model for coding and long-running professional tasks. It is built for agents that operate across entire…" }, { "rank": 7, "model": "Minimax M2.5", "provider": "MiniMax", "score": "161B", "change": -3, "summary": "MiniMax-M2.5 is a SOTA large language model designed for real-world productivity. Trained in a diverse range of complex real-world digita…", "link": "https://openrouter.ai/minimax/minimax-m2.5-20260211", "metrics": { "window": "24h", "tokens": "161B", "change": "-3%", "context": "197K" }, "secondary_en": "MiniMax", "secondary_zh": "MiniMax", "summary_zh": "MiniMax-M2.5是一款专为现实世界生产力设计的先进大型语言模型。它通过多样化的复杂现实世界数字场景训练而成…", "summary_en": "MiniMax-M2.5 is a SOTA large language model designed for real-world productivity. Trained in a diverse range of complex real-world digita…" }, { "rank": 8, "model": "Gemini 3 Flash Preview", "provider": "Google", "score": "159B", "change": 5, "summary": "Gemini 3 Flash Preview is a high speed, high value thinking model designed for agentic workflows, multi turn chat, and coding assistance.…", "link": "https://openrouter.ai/google/gemini-3-flash-preview-20251217", "metrics": { "window": "24h", "tokens": "159B", "change": "+5%", "context": "1.0M" }, "secondary_en": "Google", "secondary_zh": "Google", "summary_zh": "Gemini 3 Flash Preview是一款专为智能体工作流、多轮对话及编程辅助设计的高速、高性价比思维模型。", "summary_en": "Gemini 3 Flash Preview is a high speed, high value thinking model designed for agentic workflows, multi turn chat, and coding assistance.…" }, { "rank": 9, "model": "GLM 5 Turbo", "provider": "Z.ai", "score": "142B", "change": -1, "summary": "GLM-5 Turbo is a new model from Z.ai designed for fast inference and strong performance in agent-driven environments such as OpenClaw sce…", "link": "https://openrouter.ai/z-ai/glm-5-turbo-20260315", "metrics": { "window": "24h", "tokens": "142B", "change": "-1%", "context": "203K" }, "secondary_en": "Z.ai", "secondary_zh": "Z.ai", "summary_zh": "GLM-5 Turbo是Z.ai推出的新模型，专为快速推理和在OpenClaw等智能体驱动环境中实现强劲性能而设计…", "summary_en": "GLM-5 Turbo is a new model from Z.ai designed for fast inference and strong performance in agent-driven environments such as OpenClaw sce…" }, { "rank": 10, "model": "Gemini 2.5 Flash Lite", "provider": "Google", "score": "90.9B", "change": 5, "summary": "Gemini 2.5 Flash-Lite is a lightweight reasoning model in the Gemini 2.5 family, optimized for ultra-low latency and cost efficiency. It…", "link": "https://openrouter.ai/google/gemini-2.5-flash-lite", "metrics": { "window": "24h", "tokens": "90.9B", "change": "+5%", "context": "1.0M" }, "secondary_en": "Google", "secondary_zh": "Google", "summary_zh": "Gemini 2.5 Flash-Lite 是 Gemini 2.5 系列中的轻量级推理模型，专为超低延迟和成本效益优化。它…", "summary_en": "Gemini 2.5 Flash-Lite is a lightweight reasoning model in the Gemini 2.5 family, optimized for ultra-low latency and cost efficiency. It…" } ] }, { "id": "llm-month", "title": "LLM 月榜", "description": "查看最近一个月的模型热度与用量变化。", "category": "真实使用", "sourceLabel": "公开榜单页面与模型目录", "sourceUrl": "https://openrouter.ai/rankings?view=month", "updatedAt": "2026-03-26T07:00:14.162Z", "methodology": "直接抓取公开榜单页面中的 Top Models 区块，并补充模型上下文和说明。", "items": [ { "rank": 1, "model": "Minimax M2.5", "provider": "MiniMax", "score": "6.79T", "change": 59, "summary": "MiniMax-M2.5 is a SOTA large language model designed for real-world productivity. Trained in a diverse range of complex real-world digita…", "link": "https://openrouter.ai/minimax/minimax-m2.5-20260211", "metrics": { "window": "30d", "tokens": "6.79T", "change": "+59%", "context": "197K" }, "secondary_en": "MiniMax", "secondary_zh": "MiniMax", "summary_zh": "MiniMax-M2.5是一款专为现实世界生产力设计的先进大型语言模型。它通过多样化的复杂现实世界数字场景训练而成…", "summary_en": "MiniMax-M2.5 is a SOTA large language model designed for real-world productivity. Trained in a diverse range of complex real-world digita…" }, { "rank": 2, "model": "Step 3.5 Flash (free)", "provider": "StepFun", "score": "4.62T", "change": 498, "summary": "Step 3.5 Flash is StepFun's most capable open-source foundation model. Built on a sparse Mixture of Experts (MoE) architecture, it select…", "link": "https://openrouter.ai/stepfun/step-3.5-flash:free", "metrics": { "window": "30d", "tokens": "4.62T", "change": "+498%", "context": "256K" }, "secondary_en": "StepFun", "secondary_zh": "StepFun", "summary_zh": "Step 3.5 Flash是StepFun最强大的开源基础模型。它基于稀疏专家混合（MoE）架构构建，能够智能选择……", "summary_en": "Step 3.5 Flash is StepFun's most capable open-source foundation model. Built on a sparse Mixture of Experts (MoE) architecture, it select…" }, { "rank": 3, "model": "Gemini 3 Flash Preview", "provider": "Google", "score": "4.33T", "change": 31, "summary": "Gemini 3 Flash Preview is a high speed, high value thinking model designed for agentic workflows, multi turn chat, and coding assistance.…", "link": "https://openrouter.ai/google/gemini-3-flash-preview-20251217", "metrics": { "window": "30d", "tokens": "4.33T", "change": "+31%", "context": "1.0M" }, "secondary_en": "Google", "secondary_zh": "Google", "summary_zh": "Gemini 3 Flash Preview是一款专为智能体工作流、多轮对话及编程辅助设计的高速、高性价比思维模型。", "summary_en": "Gemini 3 Flash Preview is a high speed, high value thinking model designed for agentic workflows, multi turn chat, and coding assistance.…" }, { "rank": 4, "model": "Deepseek V3.2", "provider": "DeepSeek", "score": "4.22T", "change": 48, "summary": "DeepSeek-V3.2 is a large language model designed to harmonize high computational efficiency with strong reasoning and agentic tool-use pe…", "link": "https://openrouter.ai/deepseek/deepseek-v3.2-20251201", "metrics": { "window": "30d", "tokens": "4.22T", "change": "+48%", "context": "164K" }, "secondary_en": "DeepSeek", "secondary_zh": "DeepSeek", "summary_zh": "DeepSeek-V3.2是一款大型语言模型，旨在实现高计算效率与强大推理及智能体工具使用能力的完美结合……", "summary_en": "DeepSeek-V3.2 is a large language model designed to harmonize high computational efficiency with strong reasoning and agentic tool-use pe…" }, { "rank": 5, "model": "Claude Sonnet 4.6", "provider": "Anthropic", "score": "3.67T", "change": 885, "summary": "Sonnet 4.6 is Anthropic's most capable Sonnet-class model yet, with frontier performance across coding, agents, and professional work. It…", "link": "https://openrouter.ai/anthropic/claude-4.6-sonnet-20260217", "metrics": { "window": "30d", "tokens": "3.67T", "change": "+885%", "context": "1.0M" }, "secondary_en": "Anthropic", "secondary_zh": "Anthropic", "summary_zh": "Sonnet 4.6是Anthropic迄今为止最强大的Sonnet系列模型，在编程、智能体应用和专业工作领域均展现出前沿性能。它……", "summary_en": "Sonnet 4.6 is Anthropic's most capable Sonnet-class model yet, with frontier performance across coding, agents, and professional work. It…" }, { "rank": 6, "model": "Claude Opus 4.6", "provider": "Anthropic", "score": "3.6T", "change": 159, "summary": "Opus 4.6 is Anthropic’s strongest model for coding and long-running professional tasks. It is built for agents that operate across entire…", "link": "https://openrouter.ai/anthropic/claude-4.6-opus-20260205", "metrics": { "window": "30d", "tokens": "3.6T", "change": "+159%", "context": "1.0M" }, "secondary_en": "Anthropic", "secondary_zh": "Anthropic", "summary_zh": "Opus 4.6是Anthropic针对编程和长时间专业任务推出的最强模型。它专为能在整个...范围内运行的智能体而构建。", "summary_en": "Opus 4.6 is Anthropic’s strongest model for coding and long-running professional tasks. It is built for agents that operate across entire…" }, { "rank": 7, "model": "Mimo V2 Pro", "provider": "Xiaomi", "score": "2.97T", "change": null, "summary": "MiMo-V2-Pro is Xiaomi's flagship foundation model, featuring over 1T total parameters and a 1M context length, deeply optimized for agent…", "link": "https://openrouter.ai/xiaomi/mimo-v2-pro-20260318", "metrics": { "window": "30d", "tokens": "2.97T", "change": "new", "context": "1.0M" }, "secondary_en": "Xiaomi", "secondary_zh": "Xiaomi", "summary_zh": "MiMo-V2-Pro是小米的旗舰基础模型，拥有超过1万亿总参数和100万上下文长度，专为智能体场景深度优化……", "summary_en": "MiMo-V2-Pro is Xiaomi's flagship foundation model, featuring over 1T total parameters and a 1M context length, deeply optimized for agent…" }, { "rank": 8, "model": "Kimi K2.5 0127", "provider": "MoonshotAI", "score": "2.72T", "change": -31, "summary": "Kimi K2.5 is Moonshot AI's native multimodal model, delivering state-of-the-art visual coding capability and a self-directed agent swarm…", "link": "https://openrouter.ai/moonshotai/kimi-k2.5-0127", "metrics": { "window": "30d", "tokens": "2.72T", "change": "-31%", "context": "262K" }, "secondary_en": "MoonshotAI", "secondary_zh": "MoonshotAI", "summary_zh": "Kimi K2.5是月之暗面AI自主研发的原生多模态模型，具备顶尖的视觉编码能力与自主智能体集群…", "summary_en": "Kimi K2.5 is Moonshot AI's native multimodal model, delivering state-of-the-art visual coding capability and a self-directed agent swarm…" }, { "rank": 9, "model": "Gemini 2.5 Flash", "provider": "Google", "score": "2.35T", "change": 29, "summary": "Gemini 2.5 Flash is Google's state-of-the-art workhorse model, specifically designed for advanced reasoning, coding, mathematics, and sci…", "link": "https://openrouter.ai/google/gemini-2.5-flash", "metrics": { "window": "30d", "tokens": "2.35T", "change": "+29%", "context": "1.0M" }, "secondary_en": "Google", "secondary_zh": "Google", "summary_zh": "Gemini 2.5 Flash是谷歌推出的尖端主力模型，专为高级推理、编程、数学及科学计算等领域而设计。", "summary_en": "Gemini 2.5 Flash is Google's state-of-the-art workhorse model, specifically designed for advanced reasoning, coding, mathematics, and sci…" }, { "rank": 10, "model": "Grok 4.1 Fast", "provider": "xAI", "score": "2.33T", "change": 15, "summary": "Grok 4.1 Fast is xAI's best agentic tool calling model that shines in real-world use cases like customer support and deep research. 2M co…", "link": "https://openrouter.ai/x-ai/grok-4.1-fast", "metrics": { "window": "30d", "tokens": "2.33T", "change": "+15%", "context": "2.0M" }, "secondary_en": "xAI", "secondary_zh": "xAI", "summary_zh": "Grok 4.1 Fast是xAI旗下最出色的代理工具调用模型，在客户支持和深度研究等实际应用场景中表现卓越。", "summary_en": "Grok 4.1 Fast is xAI's best agentic tool calling model that shines in real-world use cases like customer support and deep research. 2M co…" } ] }, { "id": "apps-week", "title": "Top Apps 周榜", "description": "查看最近一周的热门应用排行。", "category": "应用生态", "sourceLabel": "公开榜单页面", "sourceUrl": "https://openrouter.ai/rankings", "updatedAt": "2026-03-26T07:00:14.163Z", "methodology": "直接抓取公开榜单页面中的 Top Apps 列表前 10 项。", "items": [ { "rank": 1, "model": "OpenClaw", "provider": "The AI that actually does things", "score": "799B", "change": 0, "summary": "The AI that actually does things", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fopenclaw.ai%2F", "metrics": { "tokens": "799B" }, "summary_zh": "真正做事的AI", "summary_en": "The AI that actually does things", "secondary_zh": "真正做事的AI", "secondary_en": "The AI that actually does things" }, { "rank": 2, "model": "Kilo Code", "provider": "AI coding agent for VS Code", "score": "212B", "change": 0, "summary": "AI coding agent for VS Code", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fkilocode.ai%2F", "metrics": { "tokens": "212B" }, "summary_zh": "VS Code AI 编码助手", "summary_en": "AI coding agent for VS Code", "secondary_zh": "VS Code AI 编码助手", "secondary_en": "AI coding agent for VS Code" }, { "rank": 3, "model": "Claude Code", "provider": "The AI for problem solvers", "score": "115B", "change": 0, "summary": "The AI for problem solvers", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fclaude.ai%2F", "metrics": { "tokens": "115B" }, "summary_zh": "问题解决者的人工智能", "summary_en": "The AI for problem solvers", "secondary_zh": "问题解决者的人工智能", "secondary_en": "The AI for problem solvers" }, { "rank": 4, "model": "Cline", "provider": "Autonomous coding agent right in your IDE", "score": "71.1B", "change": 0, "summary": "Autonomous coding agent right in your IDE", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fcline.bot%2F", "metrics": { "tokens": "71.1B" }, "summary_zh": "集成开发环境中的自主编码代理", "summary_en": "Autonomous coding agent right in your IDE", "secondary_zh": "集成开发环境中的自主编码代理", "secondary_en": "Autonomous coding agent right in your IDE" }, { "rank": 5, "model": "Descript", "provider": "AI Video & Podcast Editor", "score": "48.2B", "change": 0, "summary": "AI Video & Podcast Editor", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fdescript.com%2F", "metrics": { "tokens": "48.2B" }, "summary_zh": "AI视频与播客编辑器", "summary_en": "AI Video & Podcast Editor", "secondary_zh": "AI视频与播客编辑器", "secondary_en": "AI Video & Podcast Editor" }, { "rank": 6, "model": "Janitor AI", "provider": "Character chat and creation", "score": "27.2B", "change": 0, "summary": "Character chat and creation", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fjanitorai.com%2F", "metrics": { "tokens": "27.2B" }, "summary_zh": "角色聊天与创建", "summary_en": "Character chat and creation", "secondary_zh": "角色聊天与创建", "secondary_en": "Character chat and creation" }, { "rank": 7, "model": "ISEKAI ZERO", "provider": "AI adventures. Travel with your favorite characters", "score": "24.9B", "change": 0, "summary": "AI adventures. Travel with your favorite characters", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fwww.isekai.world%2F", "metrics": { "tokens": "24.9B" }, "summary_zh": "AI冒险之旅。与你最喜爱的角色一同旅行。", "summary_en": "AI adventures. Travel with your favorite characters", "secondary_zh": "AI冒险之旅。与你最喜爱的角色一同旅行。", "secondary_en": "AI adventures. Travel with your favorite characters" }, { "rank": 8, "model": "Roo Code", "provider": "A whole dev team of AI agents in your editor", "score": "15.3B", "change": 0, "summary": "A whole dev team of AI agents in your editor", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fgithub.com%2FRooVetGit%2FRoo-Cline", "metrics": { "tokens": "15.3B" }, "summary_zh": "编辑器中的完整AI智能体开发团队", "summary_en": "A whole dev team of AI agents in your editor", "secondary_zh": "编辑器中的完整AI智能体开发团队", "secondary_en": "A whole dev team of AI agents in your editor" }, { "rank": 9, "model": "Hermes Agent", "provider": "Control panel for AI apps", "score": "9.86B", "change": 0, "summary": "Control panel for AI apps", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fhermes-agent.nousresearch.com%2F", "metrics": { "tokens": "9.86B" }, "summary_zh": "AI应用控制面板", "summary_en": "Control panel for AI apps", "secondary_zh": "AI应用控制面板", "secondary_en": "Control panel for AI apps" } ] }, { "id": "apps-day", "title": "Top Apps 日榜", "description": "查看最近一天的热门应用排行。", "category": "应用生态", "sourceLabel": "公开榜单页面", "sourceUrl": "https://openrouter.ai/rankings?view=day", "updatedAt": "2026-03-26T07:00:14.163Z", "methodology": "直接抓取公开榜单页面中的 Top Apps 列表前 10 项。", "items": [ { "rank": 1, "model": "OpenClaw", "provider": "The AI that actually does things", "score": "799B", "change": 0, "summary": "The AI that actually does things", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fopenclaw.ai%2F", "metrics": { "tokens": "799B" }, "summary_zh": "真正做事的AI", "summary_en": "The AI that actually does things", "secondary_zh": "真正做事的AI", "secondary_en": "The AI that actually does things" }, { "rank": 2, "model": "Kilo Code", "provider": "AI coding agent for VS Code", "score": "212B", "change": 0, "summary": "AI coding agent for VS Code", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fkilocode.ai%2F", "metrics": { "tokens": "212B" }, "summary_zh": "VS Code AI 编码助手", "summary_en": "AI coding agent for VS Code", "secondary_zh": "VS Code AI 编码助手", "secondary_en": "AI coding agent for VS Code" }, { "rank": 3, "model": "Claude Code", "provider": "The AI for problem solvers", "score": "115B", "change": 0, "summary": "The AI for problem solvers", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fclaude.ai%2F", "metrics": { "tokens": "115B" }, "summary_zh": "问题解决者的人工智能", "summary_en": "The AI for problem solvers", "secondary_zh": "问题解决者的人工智能", "secondary_en": "The AI for problem solvers" }, { "rank": 4, "model": "Cline", "provider": "Autonomous coding agent right in your IDE", "score": "71.1B", "change": 0, "summary": "Autonomous coding agent right in your IDE", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fcline.bot%2F", "metrics": { "tokens": "71.1B" }, "summary_zh": "集成开发环境中的自主编码代理", "summary_en": "Autonomous coding agent right in your IDE", "secondary_zh": "集成开发环境中的自主编码代理", "secondary_en": "Autonomous coding agent right in your IDE" }, { "rank": 5, "model": "Descript", "provider": "AI Video & Podcast Editor", "score": "48.2B", "change": 0, "summary": "AI Video & Podcast Editor", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fdescript.com%2F", "metrics": { "tokens": "48.2B" }, "summary_zh": "AI视频与播客编辑器", "summary_en": "AI Video & Podcast Editor", "secondary_zh": "AI视频与播客编辑器", "secondary_en": "AI Video & Podcast Editor" }, { "rank": 6, "model": "Janitor AI", "provider": "Character chat and creation", "score": "27.2B", "change": 0, "summary": "Character chat and creation", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fjanitorai.com%2F", "metrics": { "tokens": "27.2B" }, "summary_zh": "角色聊天与创建", "summary_en": "Character chat and creation", "secondary_zh": "角色聊天与创建", "secondary_en": "Character chat and creation" }, { "rank": 7, "model": "ISEKAI ZERO", "provider": "AI adventures. Travel with your favorite characters", "score": "24.9B", "change": 0, "summary": "AI adventures. Travel with your favorite characters", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fwww.isekai.world%2F", "metrics": { "tokens": "24.9B" }, "summary_zh": "AI冒险之旅。与你最喜爱的角色一同旅行。", "summary_en": "AI adventures. Travel with your favorite characters", "secondary_zh": "AI冒险之旅。与你最喜爱的角色一同旅行。", "secondary_en": "AI adventures. Travel with your favorite characters" }, { "rank": 8, "model": "Roo Code", "provider": "A whole dev team of AI agents in your editor", "score": "15.3B", "change": 0, "summary": "A whole dev team of AI agents in your editor", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fgithub.com%2FRooVetGit%2FRoo-Cline", "metrics": { "tokens": "15.3B" }, "summary_zh": "编辑器中的完整AI智能体开发团队", "summary_en": "A whole dev team of AI agents in your editor", "secondary_zh": "编辑器中的完整AI智能体开发团队", "secondary_en": "A whole dev team of AI agents in your editor" }, { "rank": 9, "model": "Hermes Agent", "provider": "Control panel for AI apps", "score": "9.86B", "change": 0, "summary": "Control panel for AI apps", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fhermes-agent.nousresearch.com%2F", "metrics": { "tokens": "9.86B" }, "summary_zh": "AI应用控制面板", "summary_en": "Control panel for AI apps", "secondary_zh": "AI应用控制面板", "secondary_en": "Control panel for AI apps" } ] }, { "id": "apps-month", "title": "Top Apps 月榜", "description": "查看最近一个月的热门应用排行。", "category": "应用生态", "sourceLabel": "公开榜单页面", "sourceUrl": "https://openrouter.ai/rankings?view=month", "updatedAt": "2026-03-26T07:00:14.163Z", "methodology": "直接抓取公开榜单页面中的 Top Apps 列表前 10 项。", "items": [ { "rank": 1, "model": "OpenClaw", "provider": "The AI that actually does things", "score": "799B", "change": 0, "summary": "The AI that actually does things", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fopenclaw.ai%2F", "metrics": { "tokens": "799B" }, "summary_zh": "真正做事的AI", "summary_en": "The AI that actually does things", "secondary_zh": "真正做事的AI", "secondary_en": "The AI that actually does things" }, { "rank": 2, "model": "Kilo Code", "provider": "AI coding agent for VS Code", "score": "212B", "change": 0, "summary": "AI coding agent for VS Code", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fkilocode.ai%2F", "metrics": { "tokens": "212B" }, "summary_zh": "VS Code AI 编码助手", "summary_en": "AI coding agent for VS Code", "secondary_zh": "VS Code AI 编码助手", "secondary_en": "AI coding agent for VS Code" }, { "rank": 3, "model": "Claude Code", "provider": "The AI for problem solvers", "score": "115B", "change": 0, "summary": "The AI for problem solvers", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fclaude.ai%2F", "metrics": { "tokens": "115B" }, "summary_zh": "问题解决者的人工智能", "summary_en": "The AI for problem solvers", "secondary_zh": "问题解决者的人工智能", "secondary_en": "The AI for problem solvers" }, { "rank": 4, "model": "Cline", "provider": "Autonomous coding agent right in your IDE", "score": "71.1B", "change": 0, "summary": "Autonomous coding agent right in your IDE", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fcline.bot%2F", "metrics": { "tokens": "71.1B" }, "summary_zh": "集成开发环境中的自主编码代理", "summary_en": "Autonomous coding agent right in your IDE", "secondary_zh": "集成开发环境中的自主编码代理", "secondary_en": "Autonomous coding agent right in your IDE" }, { "rank": 5, "model": "Descript", "provider": "AI Video & Podcast Editor", "score": "48.2B", "change": 0, "summary": "AI Video & Podcast Editor", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fdescript.com%2F", "metrics": { "tokens": "48.2B" }, "summary_zh": "AI视频与播客编辑器", "summary_en": "AI Video & Podcast Editor", "secondary_zh": "AI视频与播客编辑器", "secondary_en": "AI Video & Podcast Editor" }, { "rank": 6, "model": "Janitor AI", "provider": "Character chat and creation", "score": "27.2B", "change": 0, "summary": "Character chat and creation", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fjanitorai.com%2F", "metrics": { "tokens": "27.2B" }, "summary_zh": "角色聊天与创建", "summary_en": "Character chat and creation", "secondary_zh": "角色聊天与创建", "secondary_en": "Character chat and creation" }, { "rank": 7, "model": "ISEKAI ZERO", "provider": "AI adventures. Travel with your favorite characters", "score": "24.9B", "change": 0, "summary": "AI adventures. Travel with your favorite characters", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fwww.isekai.world%2F", "metrics": { "tokens": "24.9B" }, "summary_zh": "AI冒险之旅。与你最喜爱的角色一同旅行。", "summary_en": "AI adventures. Travel with your favorite characters", "secondary_zh": "AI冒险之旅。与你最喜爱的角色一同旅行。", "secondary_en": "AI adventures. Travel with your favorite characters" }, { "rank": 8, "model": "Roo Code", "provider": "A whole dev team of AI agents in your editor", "score": "15.3B", "change": 0, "summary": "A whole dev team of AI agents in your editor", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fgithub.com%2FRooVetGit%2FRoo-Cline", "metrics": { "tokens": "15.3B" }, "summary_zh": "编辑器中的完整AI智能体开发团队", "summary_en": "A whole dev team of AI agents in your editor", "secondary_zh": "编辑器中的完整AI智能体开发团队", "secondary_en": "A whole dev team of AI agents in your editor" }, { "rank": 9, "model": "Hermes Agent", "provider": "Control panel for AI apps", "score": "9.86B", "change": 0, "summary": "Control panel for AI apps", "link": "https://openrouter.ai/apps?url=https%3A%2F%2Fhermes-agent.nousresearch.com%2F", "metrics": { "tokens": "9.86B" }, "summary_zh": "AI应用控制面板", "summary_en": "Control panel for AI apps", "secondary_zh": "AI应用控制面板", "secondary_en": "Control panel for AI apps" } ] }, { "id": "llm-agentic", "title": "LLM 智能体榜", "description": "查看面向智能体场景的模型表现排行。", "category": "Benchmark", "sourceLabel": "公开 benchmark 数据", "sourceUrl": "https://openrouter.ai/rankings?benchmark=agentic", "updatedAt": "2026-03-26T07:00:14.164Z", "methodology": "解析公开 benchmark 页面内嵌的数据块，提取 agentic 榜单，并补充模型元数据。", "items": [ { "rank": 1, "model": "GPT-5.4", "provider": "OpenAI", "score": "69.4", "change": 0, "summary": "GPT-5.4 is OpenAI’s latest frontier model, unifying the Codex and GPT lines into a single system. It features a 1M+ token context window…", "link": "https://openrouter.ai/openai/gpt-5.4", "metrics": { "benchmark": "agentic", "score": "69.4", "openrouter_slug": "openai/gpt-5.4", "context": "1.1M" }, "secondary_en": "OpenAI", "secondary_zh": "OpenAI", "summary_zh": "GPT-5.4是OpenAI最新的前沿模型，它将Codex和GPT系列统一整合为一个单一系统。该模型具备超过100万令牌的上下文窗口……", "summary_en": "GPT-5.4 is OpenAI’s latest frontier model, unifying the Codex and GPT lines into a single system. It features a 1M+ token context window…" }, { "rank": 2, "model": "Claude Opus 4.6", "provider": "Anthropic", "score": "67.6", "change": 0, "summary": "Opus 4.6 is Anthropic’s strongest model for coding and long-running professional tasks. It is built for agents that operate across entire…", "link": "https://openrouter.ai/anthropic/claude-opus-4.6", "metrics": { "benchmark": "agentic", "score": "67.6", "openrouter_slug": "anthropic/claude-opus-4.6", "context": "1.0M" }, "secondary_en": "Anthropic", "secondary_zh": "Anthropic", "summary_zh": "Opus 4.6是Anthropic针对编程和长时间专业任务推出的最强模型。它专为能在整个...范围内运行的智能体而构建。", "summary_en": "Opus 4.6 is Anthropic’s strongest model for coding and long-running professional tasks. It is built for agents that operate across entire…" }, { "rank": 3, "model": "GLM 5 Turbo", "provider": "Z.ai", "score": "66.1", "change": 0, "summary": "GLM-5 Turbo is a new model from Z.ai designed for fast inference and strong performance in agent-driven environments such as OpenClaw sce…", "link": "https://openrouter.ai/z-ai/glm-5-turbo", "metrics": { "benchmark": "agentic", "score": "66.1", "openrouter_slug": "z-ai/glm-5-turbo", "context": "203K" }, "secondary_en": "Z.ai", "secondary_zh": "Z.ai", "summary_zh": "GLM-5 Turbo是Z.ai推出的新模型，专为快速推理和在OpenClaw等智能体驱动环境中实现强劲性能而设计…", "summary_en": "GLM-5 Turbo is a new model from Z.ai designed for fast inference and strong performance in agent-driven environments such as OpenClaw sce…" }, { "rank": 4, "model": "GLM 5", "provider": "Z.ai", "score": "63.1", "change": 0, "summary": "GLM-5 is Z.ai’s flagship open-source foundation model engineered for complex systems design and long-horizon agent workflows. Built for e…", "link": "https://openrouter.ai/z-ai/glm-5", "metrics": { "benchmark": "agentic", "score": "63.1", "openrouter_slug": "z-ai/glm-5", "context": "80K" }, "secondary_en": "Z.ai", "secondary_zh": "Z.ai", "summary_zh": "GLM-5是Z.ai推出的旗舰开源基础模型，专为复杂系统设计与长周期智能体工作流而打造。", "summary_en": "GLM-5 is Z.ai’s flagship open-source foundation model engineered for complex systems design and long-horizon agent workflows. Built for e…" }, { "rank": 5, "model": "Claude Sonnet 4.6", "provider": "Anthropic", "score": "63", "change": 0, "summary": "Sonnet 4.6 is Anthropic's most capable Sonnet-class model yet, with frontier performance across coding, agents, and professional work. It…", "link": "https://openrouter.ai/anthropic/claude-sonnet-4.6", "metrics": { "benchmark": "agentic", "score": "63", "openrouter_slug": "anthropic/claude-sonnet-4.6", "context": "1.0M" }, "secondary_en": "Anthropic", "secondary_zh": "Anthropic", "summary_zh": "Sonnet 4.6是Anthropic迄今为止最强大的Sonnet系列模型，在编程、智能体应用和专业工作领域均展现出前沿性能。它……", "summary_en": "Sonnet 4.6 is Anthropic's most capable Sonnet-class model yet, with frontier performance across coding, agents, and professional work. It…" }, { "rank": 6, "model": "MiMo-V2-Pro", "provider": "Xiaomi", "score": "62.8", "change": 0, "summary": "MiMo-V2-Pro is Xiaomi's flagship foundation model, featuring over 1T total parameters and a 1M context length, deeply optimized for agent…", "link": "https://openrouter.ai/xiaomi/mimo-v2-pro", "metrics": { "benchmark": "agentic", "score": "62.8", "openrouter_slug": "xiaomi/mimo-v2-pro", "context": "1.0M" }, "secondary_en": "Xiaomi", "secondary_zh": "Xiaomi", "summary_zh": "MiMo-V2-Pro是小米的旗舰基础模型，拥有超过1万亿总参数和100万上下文长度，专为智能体场景深度优化……", "summary_en": "MiMo-V2-Pro is Xiaomi's flagship foundation model, featuring over 1T total parameters and a 1M context length, deeply optimized for agent…" }, { "rank": 7, "model": "GPT-5.3-Codex", "provider": "OpenAI", "score": "62.2", "change": 0, "summary": "GPT-5.3-Codex is OpenAI’s most advanced agentic coding model, combining the frontier software engineering performance of GPT-5.2-Codex wi…", "link": "https://openrouter.ai/openai/gpt-5.3-codex", "metrics": { "benchmark": "agentic", "score": "62.2", "openrouter_slug": "openai/gpt-5.3-codex", "context": "400K" }, "secondary_en": "OpenAI", "secondary_zh": "OpenAI", "summary_zh": "GPT-5.3-Codex是OpenAI最先进的代理式编码模型，它融合了GPT-5.2-Codex的前沿软件工程性能……", "summary_en": "GPT-5.3-Codex is OpenAI’s most advanced agentic coding model, combining the frontier software engineering performance of GPT-5.2-Codex wi…" }, { "rank": 8, "model": "MiniMax M2.7", "provider": "MiniMax", "score": "61.5", "change": 0, "summary": "MiniMax-M2.7 is a next-generation large language model designed for autonomous, real-world productivity and continuous improvement. Built…", "link": "https://openrouter.ai/minimax/minimax-m2.7", "metrics": { "benchmark": "agentic", "score": "61.5", "openrouter_slug": "minimax/minimax-m2.7", "context": "205K" }, "secondary_en": "MiniMax", "secondary_zh": "MiniMax", "summary_zh": "MiniMax-M2.7是新一代大型语言模型，专为自主、现实世界生产力及持续改进而设计。构建……", "summary_en": "MiniMax-M2.7 is a next-generation large language model designed for autonomous, real-world productivity and continuous improvement. Built…" }, { "rank": 9, "model": "GPT-5.2", "provider": "OpenAI", "score": "60.2", "change": 0, "summary": "GPT-5.2 is the latest frontier-grade model in the GPT-5 series, offering stronger agentic and long context perfomance compared to GPT-5.1…", "link": "https://openrouter.ai/openai/gpt-5.2", "metrics": { "benchmark": "agentic", "score": "60.2", "openrouter_slug": "openai/gpt-5.2", "context": "400K" }, "secondary_en": "OpenAI", "secondary_zh": "OpenAI", "summary_zh": "GPT-5.2是GPT-5系列中的最新前沿级模型，相比GPT-5.1，它在代理能力和长上下文处理性能上表现更加强大。", "summary_en": "GPT-5.2 is the latest frontier-grade model in the GPT-5 series, offering stronger agentic and long context perfomance compared to GPT-5.1…" }, { "rank": 10, "model": "Claude Opus 4.5", "provider": "Anthropic", "score": "59.6", "change": 0, "summary": "Claude Opus 4.5 is Anthropic’s frontier reasoning model optimized for complex software engineering, agentic workflows, and long-horizon c…", "link": "https://openrouter.ai/anthropic/claude-opus-4.5", "metrics": { "benchmark": "agentic", "score": "59.6", "openrouter_slug": "anthropic/claude-opus-4.5", "context": "200K" }, "secondary_en": "Anthropic", "secondary_zh": "Anthropic", "summary_zh": "Claude Opus 4.5是Anthropic公司推出的前沿推理模型，专为复杂软件工程、智能体工作流和长周期任务优化设计。", "summary_en": "Claude Opus 4.5 is Anthropic’s frontier reasoning model optimized for complex software engineering, agentic workflows, and long-horizon c…" } ] }, { "id": "llm-coding", "title": "LLM 编程榜", "description": "查看面向编程任务的模型表现排行。", "category": "Benchmark", "sourceLabel": "公开 benchmark 数据", "sourceUrl": "https://openrouter.ai/rankings?benchmark=coding", "updatedAt": "2026-03-26T07:00:14.164Z", "methodology": "解析公开 benchmark 页面内嵌的数据块，提取 coding 榜单，并补充模型元数据。", "items": [ { "rank": 1, "model": "GPT-5.4", "provider": "OpenAI", "score": "57.3", "change": 0, "summary": "GPT-5.4 is OpenAI’s latest frontier model, unifying the Codex and GPT lines into a single system. It features a 1M+ token context window…", "link": "https://openrouter.ai/openai/gpt-5.4", "metrics": { "benchmark": "coding", "score": "57.3", "openrouter_slug": "openai/gpt-5.4", "context": "1.1M" }, "secondary_en": "OpenAI", "secondary_zh": "OpenAI", "summary_zh": "GPT-5.4是OpenAI最新的前沿模型，它将Codex和GPT系列统一整合为一个单一系统。该模型具备超过100万令牌的上下文窗口……", "summary_en": "GPT-5.4 is OpenAI’s latest frontier model, unifying the Codex and GPT lines into a single system. It features a 1M+ token context window…" }, { "rank": 2, "model": "Gemini 3.1 Pro Preview", "provider": "Google", "score": "55.5", "change": 0, "summary": "Gemini 3.1 Pro Preview is Google’s frontier reasoning model, delivering enhanced software engineering performance, improved agentic relia…", "link": "https://openrouter.ai/google/gemini-3.1-pro-preview", "metrics": { "benchmark": "coding", "score": "55.5", "openrouter_slug": "google/gemini-3.1-pro-preview", "context": "1.0M" }, "secondary_en": "Google", "secondary_zh": "Google", "summary_zh": "Gemini 3.1 Pro Preview是谷歌的前沿推理模型，在软件工程性能上实现显著提升，增强了代理可靠性……", "summary_en": "Gemini 3.1 Pro Preview is Google’s frontier reasoning model, delivering enhanced software engineering performance, improved agentic relia…" }, { "rank": 3, "model": "GPT-5.3-Codex", "provider": "OpenAI", "score": "53.1", "change": 0, "summary": "GPT-5.3-Codex is OpenAI’s most advanced agentic coding model, combining the frontier software engineering performance of GPT-5.2-Codex wi…", "link": "https://openrouter.ai/openai/gpt-5.3-codex", "metrics": { "benchmark": "coding", "score": "53.1", "openrouter_slug": "openai/gpt-5.3-codex", "context": "400K" }, "secondary_en": "OpenAI", "secondary_zh": "OpenAI", "summary_zh": "GPT-5.3-Codex是OpenAI最先进的代理式编码模型，它融合了GPT-5.2-Codex的前沿软件工程性能……", "summary_en": "GPT-5.3-Codex is OpenAI’s most advanced agentic coding model, combining the frontier software engineering performance of GPT-5.2-Codex wi…" }, { "rank": 4, "model": "GPT-5.4 Mini", "provider": "OpenAI", "score": "51.5", "change": 0, "summary": "GPT-5.4 mini brings the core capabilities of GPT-5.4 to a faster, more efficient model optimized for high-throughput workloads. It suppor…", "link": "https://openrouter.ai/openai/gpt-5.4-mini", "metrics": { "benchmark": "coding", "score": "51.5", "openrouter_slug": "openai/gpt-5.4-mini", "context": "400K" }, "secondary_en": "OpenAI", "secondary_zh": "OpenAI", "summary_zh": "GPT-5.4 mini 将 GPT-5.4 的核心能力融入一个更快速、更高效的模型，专为高吞吐量工作负载优化。它支持…", "summary_en": "GPT-5.4 mini brings the core capabilities of GPT-5.4 to a faster, more efficient model optimized for high-throughput workloads. It suppor…" }, { "rank": 5, "model": "Claude Sonnet 4.6", "provider": "Anthropic", "score": "50.9", "change": 0, "summary": "Sonnet 4.6 is Anthropic's most capable Sonnet-class model yet, with frontier performance across coding, agents, and professional work. It…", "link": "https://openrouter.ai/anthropic/claude-sonnet-4.6", "metrics": { "benchmark": "coding", "score": "50.9", "openrouter_slug": "anthropic/claude-sonnet-4.6", "context": "1.0M" }, "secondary_en": "Anthropic", "secondary_zh": "Anthropic", "summary_zh": "Sonnet 4.6是Anthropic迄今为止最强大的Sonnet系列模型，在编程、智能体应用和专业工作领域均展现出前沿性能。它……", "summary_en": "Sonnet 4.6 is Anthropic's most capable Sonnet-class model yet, with frontier performance across coding, agents, and professional work. It…" }, { "rank": 6, "model": "GPT-5.2", "provider": "OpenAI", "score": "48.7", "change": 0, "summary": "GPT-5.2 is the latest frontier-grade model in the GPT-5 series, offering stronger agentic and long context perfomance compared to GPT-5.1…", "link": "https://openrouter.ai/openai/gpt-5.2", "metrics": { "benchmark": "coding", "score": "48.7", "openrouter_slug": "openai/gpt-5.2", "context": "400K" }, "secondary_en": "OpenAI", "secondary_zh": "OpenAI", "summary_zh": "GPT-5.2是GPT-5系列中的最新前沿级模型，相比GPT-5.1，它在代理能力和长上下文处理性能上表现更加强大。", "summary_en": "GPT-5.2 is the latest frontier-grade model in the GPT-5 series, offering stronger agentic and long context perfomance compared to GPT-5.1…" }, { "rank": 7, "model": "Claude Opus 4.6", "provider": "Anthropic", "score": "48.1", "change": 0, "summary": "Opus 4.6 is Anthropic’s strongest model for coding and long-running professional tasks. It is built for agents that operate across entire…", "link": "https://openrouter.ai/anthropic/claude-opus-4.6", "metrics": { "benchmark": "coding", "score": "48.1", "openrouter_slug": "anthropic/claude-opus-4.6", "context": "1.0M" }, "secondary_en": "Anthropic", "secondary_zh": "Anthropic", "summary_zh": "Opus 4.6是Anthropic针对编程和长时间专业任务推出的最强模型。它专为能在整个...范围内运行的智能体而构建。", "summary_en": "Opus 4.6 is Anthropic’s strongest model for coding and long-running professional tasks. It is built for agents that operate across entire…" }, { "rank": 8, "model": "Claude Opus 4.5", "provider": "Anthropic", "score": "47.8", "change": 0, "summary": "Claude Opus 4.5 is Anthropic’s frontier reasoning model optimized for complex software engineering, agentic workflows, and long-horizon c…", "link": "https://openrouter.ai/anthropic/claude-opus-4.5", "metrics": { "benchmark": "coding", "score": "47.8", "openrouter_slug": "anthropic/claude-opus-4.5", "context": "200K" }, "secondary_en": "Anthropic", "secondary_zh": "Anthropic", "summary_zh": "Claude Opus 4.5是Anthropic公司推出的前沿推理模型，专为复杂软件工程、智能体工作流和长周期任务优化设计。", "summary_en": "Claude Opus 4.5 is Anthropic’s frontier reasoning model optimized for complex software engineering, agentic workflows, and long-horizon c…" }, { "rank": 9, "model": "Gemini 2.5 Pro", "provider": "Google", "score": "46.7", "change": 0, "summary": "Gemini 2.5 Pro is Google’s state-of-the-art AI model designed for advanced reasoning, coding, mathematics, and scientific tasks. It emplo…", "link": "https://openrouter.ai/google/gemini-2.5-pro-exp-03-25", "metrics": { "benchmark": "coding", "score": "46.7", "openrouter_slug": "google/gemini-2.5-pro-exp-03-25", "context": "1.0M" }, "secondary_en": "Google", "secondary_zh": "Google", "summary_zh": "Gemini 2.5 Pro是谷歌推出的尖端人工智能模型，专为高级推理、编程、数学及科学任务而设计。它采用…", "summary_en": "Gemini 2.5 Pro is Google’s state-of-the-art AI model designed for advanced reasoning, coding, mathematics, and scientific tasks. It emplo…" }, { "rank": 10, "model": "Gemini 3 Pro Preview (high)", "provider": "Google", "score": "46.5", "change": 0, "summary": "Coding benchmark score.", "link": "https://openrouter.ai/google/gemini-3-pro-preview", "metrics": { "benchmark": "coding", "score": "46.5", "openrouter_slug": "google/gemini-3-pro-preview", "context": "-" }, "secondary_en": "Google", "secondary_zh": "Google", "summary_zh": "编码基准分数。", "summary_en": "Coding benchmark score." } ] }, { "id": "openclaw-success", "title": "OpenClaw 分数榜", "description": "查看 OpenClaw 标准任务的分数排行。", "category": "OpenClaw Benchmark", "sourceLabel": "公开任务榜单", "sourceUrl": "https://pinchbench.com/", "updatedAt": "2026-03-20T17:21:00.000Z", "methodology": "解析页面内嵌 entries 数据，按 success 视图生成 OpenClaw 专项榜单。", "items": [ { "rank": 1, "model": "Claude Opus 4.6", "provider": "Anthropic", "score": "82.30%", "change": 0, "summary": "OpenClaw 标准任务平均得分 82.30%，成功率 87.45%。", "link": "", "metrics": { "success_rate": "87.45%", "submissions": "17", "updated_at": "2026-03-21T00:44:45Z", "score": "82.30%", "best_time": "1032.51s", "best_cost": "$2.43" }, "secondary_en": "Anthropic", "secondary_zh": "Anthropic", "summary_zh": "OpenClaw 标准任务平均得分 82.30%，成功率 87.45%。", "summary_en": "OpenClaw standard task average score 82.30%, success rate 87.45%." }, { "rank": 2, "model": "Mimo V2 Omni", "provider": "Xiaomi", "score": "81.82%", "change": 0, "summary": "OpenClaw 标准任务平均得分 81.82%，成功率 85.61%。", "link": "", "metrics": { "success_rate": "85.61%", "submissions": "6", "updated_at": "2026-03-21T00:31:52Z", "score": "81.82%", "best_time": "708.06s", "best_cost": "$0.00" }, "secondary_en": "Xiaomi", "secondary_zh": "Xiaomi", "summary_zh": "OpenClaw 标准任务平均得分 81.82%，成功率 85.61%。", "summary_en": "OpenClaw standard task average score 81.82%, success rate 85.61%." }, { "rank": 3, "model": "Minimax M2.7", "provider": "Minimax", "score": "81.78%", "change": 0, "summary": "OpenClaw 标准任务平均得分 81.78%，成功率 87.13%。", "link": "", "metrics": { "success_rate": "87.13%", "submissions": "9", "updated_at": "2026-03-21T00:41:41Z", "score": "81.78%", "best_time": "1071.26s", "best_cost": "$0.00" }, "secondary_en": "Minimax", "secondary_zh": "Minimax", "summary_zh": "OpenClaw 标准任务平均得分 81.78%，成功率 87.13%。", "summary_en": "OpenClaw standard task average score 81.78%, success rate 87.13%." }, { "rank": 4, "model": "Gpt 5.4", "provider": "Openai", "score": "81.57%", "change": 0, "summary": "OpenClaw 标准任务平均得分 81.57%，成功率 90.47%。", "link": "", "metrics": { "success_rate": "90.47%", "submissions": "15", "updated_at": "2026-03-21T00:45:37Z", "score": "81.57%", "best_time": "958.63s", "best_cost": "$1.34" }, "secondary_en": "Openai", "secondary_zh": "Openai", "summary_zh": "OpenClaw 标准任务平均得分 81.57%，成功率 90.47%。", "summary_en": "OpenClaw standard task average score 81.57%, success rate 90.47%." }, { "rank": 5, "model": "Glm 5 Turbo", "provider": "Z Ai", "score": "81.57%", "change": 0, "summary": "OpenClaw 标准任务平均得分 81.57%，成功率 86.54%。", "link": "", "metrics": { "success_rate": "86.54%", "submissions": "9", "updated_at": "2026-03-21T00:50:48Z", "score": "81.57%", "best_time": "984.42s", "best_cost": "$0.00" }, "secondary_en": "Z Ai", "secondary_zh": "Z Ai", "summary_zh": "OpenClaw 标准任务平均得分 81.57%，成功率 86.54%。", "summary_en": "OpenClaw standard task average score 81.57%, success rate 86.54%." }, { "rank": 6, "model": "Mimo V2 Pro", "provider": "Xiaomi", "score": "80.97%", "change": 0, "summary": "OpenClaw 标准任务平均得分 80.97%，成功率 83.95%。", "link": "", "metrics": { "success_rate": "83.95%", "submissions": "13", "updated_at": "2026-03-21T01:07:20Z", "score": "80.97%", "best_time": "904.74s", "best_cost": "$0.00" }, "secondary_en": "Xiaomi", "secondary_zh": "Xiaomi", "summary_zh": "OpenClaw 标准任务平均得分 80.97%，成功率 83.95%。", "summary_en": "OpenClaw standard task average score 80.97%, success rate 83.95%." }, { "rank": 7, "model": "Qwen3.5 122b A10b", "provider": "Qwen", "score": "80.80%", "change": 0, "summary": "OpenClaw 标准任务平均得分 80.80%，成功率 85.48%。", "link": "", "metrics": { "success_rate": "85.48%", "submissions": "13", "updated_at": "2026-03-19T03:48:17Z", "score": "80.80%", "best_time": "677.34s", "best_cost": "$0.43" }, "secondary_en": "Qwen", "secondary_zh": "Qwen", "summary_zh": "OpenClaw 标准任务平均得分 80.80%，成功率 85.48%。", "summary_en": "OpenClaw standard task average score 80.80%, success rate 85.48%." }, { "rank": 8, "model": "Claude Sonnet 4.5", "provider": "Anthropic", "score": "80.67%", "change": 0, "summary": "OpenClaw 标准任务平均得分 80.67%，成功率 88.23%。", "link": "", "metrics": { "success_rate": "88.23%", "submissions": "17", "updated_at": "2026-03-21T00:41:21Z", "score": "80.67%", "best_time": "906.60s", "best_cost": "$2.14" }, "secondary_en": "Anthropic", "secondary_zh": "Anthropic", "summary_zh": "OpenClaw 标准任务平均得分 80.67%，成功率 88.23%。", "summary_en": "OpenClaw standard task average score 80.67%, success rate 88.23%." }, { "rank": 9, "model": "Claude Sonnet 4", "provider": "Anthropic", "score": "80.47%", "change": 0, "summary": "OpenClaw 标准任务平均得分 80.47%，成功率 80.47%。", "link": "", "metrics": { "success_rate": "80.47%", "submissions": "1", "updated_at": "2026-03-09T18:55:29Z", "score": "80.47%", "best_time": "932.29s", "best_cost": "$2.87" }, "secondary_en": "Anthropic", "secondary_zh": "Anthropic", "summary_zh": "OpenClaw 标准任务平均得分 80.47%，成功率 80.47%。", "summary_en": "OpenClaw standard task average score 80.47%, success rate 80.47%." }, { "rank": 10, "model": "Qwen3.5 397b A17b", "provider": "Qwen", "score": "80.45%", "change": 0, "summary": "OpenClaw 标准任务平均得分 80.45%，成功率 89.14%。", "link": "", "metrics": { "success_rate": "89.14%", "submissions": "9", "updated_at": "2026-03-19T04:03:18Z", "score": "80.45%", "best_time": "885.84s", "best_cost": "$0.75" }, "secondary_en": "Qwen", "secondary_zh": "Qwen", "summary_zh": "OpenClaw 标准任务平均得分 80.45%，成功率 89.14%。", "summary_en": "OpenClaw standard task average score 80.45%, success rate 89.14%." } ] }, { "id": "openclaw-speed", "title": "OpenClaw 速度榜", "description": "查看 OpenClaw 标准任务的速度排行。", "category": "OpenClaw Benchmark", "sourceLabel": "公开任务榜单", "sourceUrl": "https://pinchbench.com/?view=speed", "updatedAt": "2026-03-20T17:21:00.000Z", "methodology": "解析页面内嵌 entries 数据，按 speed 视图生成 OpenClaw 专项榜单。", "items": [ { "rank": 1, "model": "Gpt Oss 20b", "provider": "Openai", "score": "345.05s", "change": 0, "summary": "OpenClaw 任务最佳耗时 345.05s，成功率 65.99%。", "link": "", "metrics": { "success_rate": "65.99%", "submissions": "11", "updated_at": "2026-03-21T01:01:17Z", "best_time": "345.05s", "avg_time": "502.80s" }, "secondary_en": "Openai", "secondary_zh": "Openai", "summary_zh": "OpenClaw 任务最佳耗时 345.05s，成功率 65.99%。", "summary_en": "OpenClaw task best time 345.05s, success rate 65.99%." }, { "rank": 2, "model": "Mercury 2", "provider": "Inception", "score": "395.79s", "change": 0, "summary": "OpenClaw 任务最佳耗时 395.79s，成功率 78.01%。", "link": "", "metrics": { "success_rate": "78.01%", "submissions": "7", "updated_at": "2026-03-20T00:59:38Z", "best_time": "395.79s", "avg_time": "503.44s" }, "secondary_en": "Inception", "secondary_zh": "Inception", "summary_zh": "OpenClaw 任务最佳耗时 395.79s，成功率 78.01%。", "summary_en": "OpenClaw task best time 395.79s, success rate 78.01%." }, { "rank": 3, "model": "Deepseek V3.2", "provider": "Deepseek", "score": "398.88s", "change": 0, "summary": "OpenClaw 任务最佳耗时 398.88s，成功率 84.30%。", "link": "", "metrics": { "success_rate": "84.30%", "submissions": "15", "updated_at": "2026-03-21T01:18:15Z", "best_time": "398.88s", "avg_time": "2913.25s" }, "secondary_en": "Deepseek", "secondary_zh": "Deepseek", "summary_zh": "OpenClaw 任务最佳耗时 398.88s，成功率 84.30%。", "summary_en": "OpenClaw task best time 398.88s, success rate 84.30%." }, { "rank": 4, "model": "Mimo V2 Flash", "provider": "Xiaomi", "score": "401.91s", "change": 0, "summary": "OpenClaw 任务最佳耗时 401.91s，成功率 81.51%。", "link": "", "metrics": { "success_rate": "81.51%", "submissions": "7", "updated_at": "2026-03-21T00:34:20Z", "best_time": "401.91s", "avg_time": "1356.78s" }, "secondary_en": "Xiaomi", "secondary_zh": "Xiaomi", "summary_zh": "OpenClaw 任务最佳耗时 401.91s，成功率 81.51%。", "summary_en": "OpenClaw task best time 401.91s, success rate 81.51%." }, { "rank": 5, "model": "Gpt 4o", "provider": "Openai", "score": "445.60s", "change": 0, "summary": "OpenClaw 任务最佳耗时 445.60s，成功率 71.14%。", "link": "", "metrics": { "success_rate": "71.14%", "submissions": "16", "updated_at": "2026-03-21T00:25:43Z", "best_time": "445.60s", "avg_time": "601.26s" }, "secondary_en": "Openai", "secondary_zh": "Openai", "summary_zh": "OpenClaw 任务最佳耗时 445.60s，成功率 71.14%。", "summary_en": "OpenClaw task best time 445.60s, success rate 71.14%." }, { "rank": 6, "model": "Gemini 3 Flash Preview", "provider": "Google", "score": "545.20s", "change": 0, "summary": "OpenClaw 任务最佳耗时 545.20s，成功率 85.22%。", "link": "", "metrics": { "success_rate": "85.22%", "submissions": "16", "updated_at": "2026-03-21T00:32:37Z", "best_time": "545.20s", "avg_time": "784.84s" }, "secondary_en": "Google", "secondary_zh": "Google", "summary_zh": "OpenClaw 任务最佳耗时 545.20s，成功率 85.22%。", "summary_en": "OpenClaw task best time 545.20s, success rate 85.22%." }, { "rank": 7, "model": "Gpt 5 Mini", "provider": "Openai", "score": "553.55s", "change": 0, "summary": "OpenClaw 任务最佳耗时 553.55s，成功率 80.29%。", "link": "", "metrics": { "success_rate": "80.29%", "submissions": "15", "updated_at": "2026-03-21T00:26:59Z", "best_time": "553.55s", "avg_time": "860.93s" }, "secondary_en": "Openai", "secondary_zh": "Openai", "summary_zh": "OpenClaw 任务最佳耗时 553.55s，成功率 80.29%。", "summary_en": "OpenClaw task best time 553.55s, success rate 80.29%." }, { "rank": 8, "model": "Gpt 4o Mini", "provider": "Openai", "score": "579.34s", "change": 0, "summary": "OpenClaw 任务最佳耗时 579.34s，成功率 75.05%。", "link": "", "metrics": { "success_rate": "75.05%", "submissions": "15", "updated_at": "2026-03-21T00:33:57Z", "best_time": "579.34s", "avg_time": "975.02s" }, "secondary_en": "Openai", "secondary_zh": "Openai", "summary_zh": "OpenClaw 任务最佳耗时 579.34s，成功率 75.05%。", "summary_en": "OpenClaw task best time 579.34s, success rate 75.05%." }, { "rank": 9, "model": "Gpt 5 Nano", "provider": "Openai", "score": "620.68s", "change": 0, "summary": "OpenClaw 任务最佳耗时 620.68s，成功率 68.78%。", "link": "", "metrics": { "success_rate": "68.78%", "submissions": "15", "updated_at": "2026-03-21T00:28:58Z", "best_time": "620.68s", "avg_time": "745.92s" }, "secondary_en": "Openai", "secondary_zh": "Openai", "summary_zh": "OpenClaw 任务最佳耗时 620.68s，成功率 68.78%。", "summary_en": "OpenClaw task best time 620.68s, success rate 68.78%." }, { "rank": 10, "model": "Qwen3.5 35b A3b", "provider": "Qwen", "score": "622.13s", "change": 0, "summary": "OpenClaw 任务最佳耗时 622.13s，成功率 78.35%。", "link": "", "metrics": { "success_rate": "78.35%", "submissions": "11", "updated_at": "2026-03-18T18:29:07Z", "best_time": "622.13s", "avg_time": "894.58s" }, "secondary_en": "Qwen", "secondary_zh": "Qwen", "summary_zh": "OpenClaw 任务最佳耗时 622.13s，成功率 78.35%。", "summary_en": "OpenClaw task best time 622.13s, success rate 78.35%." } ] }, { "id": "openclaw-cost", "title": "OpenClaw 成本榜", "description": "查看 OpenClaw 标准任务的成本排行。", "category": "OpenClaw Benchmark", "sourceLabel": "公开任务榜单", "sourceUrl": "https://pinchbench.com/?view=cost", "updatedAt": "2026-03-20T17:21:00.000Z", "methodology": "解析页面内嵌 entries 数据，按 cost 视图生成 OpenClaw 专项榜单。", "items": [ { "rank": 1, "model": "Gpt Oss 120b", "provider": "Openai", "score": "$0.03", "change": 0, "summary": "OpenClaw 任务最佳成本 $0.03，成功率 60.61%。", "link": "", "metrics": { "success_rate": "60.61%", "submissions": "7", "updated_at": "2026-03-21T00:22:41Z", "best_cost": "$0.03", "avg_cost": "$0.04" }, "secondary_en": "Openai", "secondary_zh": "Openai", "summary_zh": "OpenClaw 任务最佳成本 $0.03，成功率 60.61%。", "summary_en": "OpenClaw task best cost $0.03, success rate 60.61%." }, { "rank": 2, "model": "Qwen 2.5 7b Instruct", "provider": "Qwen", "score": "$0.06", "change": 0, "summary": "OpenClaw 任务最佳成本 $0.06，成功率 40.29%。", "link": "", "metrics": { "success_rate": "40.29%", "submissions": "12", "updated_at": "2026-03-19T03:57:30Z", "best_cost": "$0.06", "avg_cost": "$0.08" }, "secondary_en": "Qwen", "secondary_zh": "Qwen", "summary_zh": "OpenClaw 任务最佳成本 $0.06，成功率 40.29%。", "summary_en": "OpenClaw task best cost $0.06, success rate 40.29%." }, { "rank": 3, "model": "Qwen3.5 35b A3b", "provider": "Qwen", "score": "$0.07", "change": 0, "summary": "OpenClaw 任务最佳成本 $0.07，成功率 78.35%。", "link": "", "metrics": { "success_rate": "78.35%", "submissions": "11", "updated_at": "2026-03-18T18:29:07Z", "best_cost": "$0.07", "avg_cost": "$0.27" }, "secondary_en": "Qwen", "secondary_zh": "Qwen", "summary_zh": "OpenClaw 任务最佳成本 $0.07，成功率 78.35%。", "summary_en": "OpenClaw task best cost $0.07, success rate 78.35%." }, { "rank": 4, "model": "Mercury 2", "provider": "Inception", "score": "$0.07", "change": 0, "summary": "OpenClaw 任务最佳成本 $0.07，成功率 78.01%。", "link": "", "metrics": { "success_rate": "78.01%", "submissions": "7", "updated_at": "2026-03-20T00:59:38Z", "best_cost": "$0.07", "avg_cost": "$0.11" }, "secondary_en": "Inception", "secondary_zh": "Inception", "summary_zh": "OpenClaw 任务最佳成本 $0.07，成功率 78.01%。", "summary_en": "OpenClaw task best cost $0.07, success rate 78.01%." }, { "rank": 5, "model": "Deepseek Chat", "provider": "Deepseek", "score": "$0.08", "change": 0, "summary": "OpenClaw 任务最佳成本 $0.08，成功率 71.69%。", "link": "", "metrics": { "success_rate": "71.69%", "submissions": "16", "updated_at": "2026-03-21T00:35:54Z", "best_cost": "$0.08", "avg_cost": "$0.19" }, "secondary_en": "Deepseek", "secondary_zh": "Deepseek", "summary_zh": "OpenClaw 任务最佳成本 $0.08，成功率 71.69%。", "summary_en": "OpenClaw task best cost $0.08, success rate 71.69%." }, { "rank": 6, "model": "Minimax M2.1", "provider": "Minimax", "score": "$0.11", "change": 0, "summary": "OpenClaw 任务最佳成本 $0.11，成功率 84.27%。", "link": "", "metrics": { "success_rate": "84.27%", "submissions": "21", "updated_at": "2026-03-21T00:46:42Z", "best_cost": "$0.11", "avg_cost": "$0.15" }, "secondary_en": "Minimax", "secondary_zh": "Minimax", "summary_zh": "OpenClaw 任务最佳成本 $0.11，成功率 84.27%。", "summary_en": "OpenClaw task best cost $0.11, success rate 84.27%." }, { "rank": 7, "model": "Glm 4.5 Air", "provider": "Z Ai", "score": "$0.11", "change": 0, "summary": "OpenClaw 任务最佳成本 $0.11，成功率 85.71%。", "link": "", "metrics": { "success_rate": "85.71%", "submissions": "15", "updated_at": "2026-03-21T00:41:09Z", "best_cost": "$0.11", "avg_cost": "$0.15" }, "secondary_en": "Z Ai", "secondary_zh": "Z Ai", "summary_zh": "OpenClaw 任务最佳成本 $0.11，成功率 85.71%。", "summary_en": "OpenClaw task best cost $0.11, success rate 85.71%." }, { "rank": 8, "model": "Minimax M2.5", "provider": "Minimax", "score": "$0.14", "change": 0, "summary": "OpenClaw 任务最佳成本 $0.14，成功率 87.82%。", "link": "", "metrics": { "success_rate": "87.82%", "submissions": "15", "updated_at": "2026-03-21T00:38:45Z", "best_cost": "$0.14", "avg_cost": "$0.18" }, "secondary_en": "Minimax", "secondary_zh": "Minimax", "summary_zh": "OpenClaw 任务最佳成本 $0.14，成功率 87.82%。", "summary_en": "OpenClaw task best cost $0.14, success rate 87.82%." }, { "rank": 9, "model": "Gemini 2.5 Flash", "provider": "Google", "score": "$0.17", "change": 0, "summary": "OpenClaw 任务最佳成本 $0.17，成功率 70.73%。", "link": "", "metrics": { "success_rate": "70.73%", "submissions": "16", "updated_at": "2026-03-21T01:15:21Z", "best_cost": "$0.17", "avg_cost": "$0.23" }, "secondary_en": "Google", "secondary_zh": "Google", "summary_zh": "OpenClaw 任务最佳成本 $0.17，成功率 70.73%。", "summary_en": "OpenClaw task optimal cost $0.17, success rate 70.73%." }, { "rank": 10, "model": "Llama 4 Maverick", "provider": "Meta Llama", "score": "$0.20", "change": 0, "summary": "OpenClaw 任务最佳成本 $0.20，成功率 46.12%。", "link": "", "metrics": { "success_rate": "46.12%", "submissions": "9", "updated_at": "2026-03-19T03:40:53Z", "best_cost": "$0.20", "avg_cost": "$0.37" }, "secondary_en": "Meta Llama", "secondary_zh": "Meta Llama", "summary_zh": "OpenClaw 任务最佳成本 $0.20，成功率 46.12%。", "summary_en": "OpenClaw task best cost $0.20, success rate 46.12%." } ] } ] }