CLSSAI 独立产品正在逐项复刻与接入真实能力。查看进度
Cclssai

MODEL INTELLIGENCE

AI model rankings

用同一页面比较模型热度、质量、能力、性能与生态信号。公开 API 只返回顺序时，我们只发布顺序；需要认证或 CLSSAI 原生聚合的数据保持明确 Pending。

Compare models Official reference
13 sections · 8/8 public orderings available · 5 authenticated sources pending338 exact CLSSAI directory IDs can receive a Live badge; all other ranked models remain external OpenRouter previews.

POPULARITY

Top Models

按最近一周处理 token 的公开服务端顺序浏览当前模型。

Public ordering
MetricWeekly token orderWindowRolling 7 days · ordinal result onlyTemporary official API
  1. #1DDeepSeek: DeepSeek V4 Flash 0423deepseek/deepseek-v4-flashWeekly orderCLSSAI live
  2. #2XXiaomi: MiMo-V2.5xiaomi/mimo-v2.5Weekly orderCLSSAI live
  3. #3TTencent: Hy3tencent/hy3Weekly orderCLSSAI live
  4. #4DDeepSeek: DeepSeek V4 Prodeepseek/deepseek-v4-proWeekly orderCLSSAI live
  5. #5ZZ.ai: GLM 5.2z-ai/glm-5.2Weekly orderCLSSAI live
  6. #6DDeepSeek: DeepSeek V4 Flash 0731deepseek/deepseek-v4-flash-0731Weekly orderCLSSAI live
  7. #7OOpenAI: GPT-5.6 Lunaopenai/gpt-5.6-lunaWeekly orderCLSSAI live
  8. #8NNVIDIA: Nemotron 3 Ultra (free)nvidia/nemotron-3-ultra-550b-a55b:freeWeekly orderCLSSAI live
  9. #9MMiniMax: MiniMax M3minimax/minimax-m3Weekly orderCLSSAI live
  10. #10SStepFun: Step 3.7 Flashstepfun/step-3.7-flashWeekly orderCLSSAI live

TABLE

Leaderboard

用可扫描表格展示与 Top Models 相同的公开周榜顺序与模型元数据。

Public ordering
MetricWeekly token order · table viewWindowRolling 7 days · same source as Top ModelsTemporary official API
RankModelStatusContextInput → output
#1DDeepSeek: DeepSeek V4 Flash 0423deepseek/deepseek-v4-flashCLSSAI live1.05Mtext → text
#2XXiaomi: MiMo-V2.5xiaomi/mimo-v2.5CLSSAI live1.05Mtext, audio, image, video → text
#3TTencent: Hy3tencent/hy3CLSSAI live262.14Ktext → text
#4DDeepSeek: DeepSeek V4 Prodeepseek/deepseek-v4-proCLSSAI live1.05Mtext → text
#5ZZ.ai: GLM 5.2z-ai/glm-5.2CLSSAI live1.05Mtext → text
#6DDeepSeek: DeepSeek V4 Flash 0731deepseek/deepseek-v4-flash-0731CLSSAI live1.05Mtext → text
#7OOpenAI: GPT-5.6 Lunaopenai/gpt-5.6-lunaCLSSAI live1.05Mfile, image, text → text
#8NNVIDIA: Nemotron 3 Ultra (free)nvidia/nemotron-3-ultra-550b-a55b:freeCLSSAI live1Mtext → text
#9MMiniMax: MiniMax M3minimax/minimax-m3CLSSAI live1.05Mtext, image, video → text
#10SStepFun: Step 3.7 Flashstepfun/step-3.7-flashCLSSAI live262.14Ktext, image, video → text
#11MMoonshotAI: Kimi K3moonshotai/kimi-k3CLSSAI live1.05Mtext, image → text
#12ILing-3.0-flash (free)inclusionai/ling-3.0-flash:freeCLSSAI live262.14Ktext → text
#13PPoolside: Laguna S 2.1 (free)poolside/laguna-s-2.1:freeCLSSAI live262.14Ktext → text
#14AClaude Opus 5anthropic/claude-opus-5CLSSAI live1Mtext, image, file → text
#15AAnthropic: Claude Sonnet 5anthropic/claude-sonnet-5CLSSAI live1Mtext, image, file → text

TASK

Top models by task

按任务分类比较模型用量或消费，同时区分精确统计与抽样估算。

Pending

Verified dataset required

Authenticated task/category dataset with sampling disclosure. 此处不使用静态模型、推测份额或合成趋势填充版面。

Track the native implementation

COST

Cost per session

按会话比较实际结算成本，并说明样本、币种、异常值与中位数口径。

Pending

Verified dataset required

Session-level usage and official-price settlement aggregation. 此处不使用静态模型、推测份额或合成趋势填充版面。

Track the native implementation

USAGE

Market Share

按可核验 token 用量比较模型份额，并保留时间窗口、样本与更新时间。

Pending

Verified dataset required

OpenRouter rankings-daily authentication or native CLSSAI usage aggregation. 此处不使用静态模型、推测份额或合成趋势填充版面。

Track the native implementation

QUALITY

Benchmarks

按公开 intelligence 排序信号浏览模型；当前只发布顺序，不展示未返回的分数。

Public ordering
MetricArtificial Analysis intelligence order via OpenRouterWindowCurrent benchmark ordering · score not returned in the parsed responseTemporary official API
  1. #1AClaude Opus 5anthropic/claude-opus-5Intelligence orderCLSSAI live
  2. #2AAnthropic: Claude Fable 5anthropic/claude-fable-5Intelligence orderCLSSAI live
  3. #3OOpenAI: GPT-5.6 Solopenai/gpt-5.6-solIntelligence orderCLSSAI live
  4. #4MMoonshotAI: Kimi K3moonshotai/kimi-k3Intelligence orderCLSSAI live
  5. #5AAnthropic: Claude Opus 4.8anthropic/claude-opus-4.8Intelligence orderCLSSAI live
  6. #6OOpenAI: GPT-5.6 Terraopenai/gpt-5.6-terraIntelligence orderCLSSAI live
  7. #7OOpenAI: GPT-5.5openai/gpt-5.5Intelligence orderCLSSAI live
  8. #8XxAI: Grok 4.5x-ai/grok-4.5Intelligence orderCLSSAI live
  9. #9AAnthropic: Claude Opus 4.7anthropic/claude-opus-4.7Intelligence orderCLSSAI live
  10. #10AAnthropic: Claude Sonnet 5anthropic/claude-sonnet-5Intelligence orderCLSSAI live

PERFORMANCE

Fastest models

使用公开吞吐量排序浏览模型；当前响应只授权顺序，不授权伪造 tok/s。

Public ordering
MetricP50 throughput routing orderWindowCurrent routing heuristic · numeric throughput not returnedTemporary official API
  1. #1MMorph: Morph V3 Largemorph/morph-v3-largeThroughput orderCLSSAI live
  2. #2MMorph: Morph V3 Fastmorph/morph-v3-fastThroughput orderCLSSAI live
  3. #3XxAI: Grok 4.20 Multi-Agentx-ai/grok-4.20-multi-agentThroughput orderCLSSAI live
  4. #4GGoogle: Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image)google/gemini-3.1-flash-lite-imageThroughput orderCLSSAI live
  5. #5OOpenAI: gpt-oss-safeguard-20bopenai/gpt-oss-safeguard-20bThroughput orderCLSSAI live
  6. #6IInception: Mercury 2inception/mercury-2Throughput orderCLSSAI live
  7. #7TThinking Machines: Inkling Smallthinkingmachines/inkling-smallThroughput orderCLSSAI live
  8. #8GGoogle: Nano Banana (Gemini 2.5 Flash Image)google/gemini-2.5-flash-imageThroughput orderCLSSAI live
  9. #9AArcee AI: Trinity Large Thinkingarcee-ai/trinity-large-thinkingThroughput orderCLSSAI live
  10. #10QQwen: Qwen3 30B A3B Thinking 2507qwen/qwen3-30b-a3b-thinking-2507Throughput orderCLSSAI live

LANGUAGE

Languages

按自然语言与编程语言比较模型使用情况，并显著披露抽样估算。

Pending

Verified dataset required

Authenticated language ranking dataset with estimate methodology. 此处不使用静态模型、推测份额或合成趋势填充版面。

Track the native implementation

TASK

Programming

查看 OpenRouter programming 分类中按最近一周 token 处理量排列的模型。

Public ordering
MetricProgramming category · weekly token orderWindowRolling 7 days · ordinal result onlyTemporary official API
  1. #1DDeepSeek: DeepSeek V4 Flash 0423deepseek/deepseek-v4-flashWeekly orderCLSSAI live
  2. #2XXiaomi: MiMo-V2.5xiaomi/mimo-v2.5Weekly orderCLSSAI live
  3. #3TTencent: Hy3tencent/hy3Weekly orderCLSSAI live
  4. #4DDeepSeek: DeepSeek V4 Prodeepseek/deepseek-v4-proWeekly orderCLSSAI live
  5. #5ZZ.ai: GLM 5.2z-ai/glm-5.2Weekly orderCLSSAI live
  6. #6DDeepSeek: DeepSeek V4 Flash 0731deepseek/deepseek-v4-flash-0731Weekly orderCLSSAI live
  7. #7OOpenAI: GPT-5.6 Lunaopenai/gpt-5.6-lunaWeekly orderCLSSAI live
  8. #8NNVIDIA: Nemotron 3 Ultra (free)nvidia/nemotron-3-ultra-550b-a55b:freeWeekly orderCLSSAI live
  9. #9MMiniMax: MiniMax M3minimax/minimax-m3Weekly orderCLSSAI live
  10. #10SStepFun: Step 3.7 Flashstepfun/step-3.7-flashWeekly orderCLSSAI live

CAPABILITY

Context Length

按目录声明的最大上下文长度从高到低比较模型。

Public ordering
MetricDeclared context lengthWindowCurrent model directory · descendingTemporary official API
  1. #1OAuto Routeropenrouter/auto2M contextCLSSAI live
  2. #2OAuto Router (Beta)openrouter/auto-beta2M contextCLSSAI live
  3. #3OPareto Code Routeropenrouter/pareto-code2M contextCLSSAI live
  4. #4XxAI: Grok 4.20x-ai/grok-4.202M contextCLSSAI live
  5. #5XxAI: Grok 4.20 Multi-Agentx-ai/grok-4.20-multi-agent2M contextCLSSAI live
  6. #6MMeta: Llama 4 Scoutmeta-llama/llama-4-scout1.31M contextCLSSAI live
  7. #7~OpenAI GPT Latest~openai/gpt-latest1.05M contextCLSSAI live
  8. #8OOpenAI: GPT-5.4openai/gpt-5.41.05M contextCLSSAI live
  9. #9OOpenAI: GPT-5.4 Proopenai/gpt-5.4-pro1.05M contextCLSSAI live
  10. #10OOpenAI: GPT-5.5openai/gpt-5.51.05M contextCLSSAI live

CAPABILITY

Tool Calls

在公开周榜顺序中精确筛选声明支持 tools 参数的模型。

Public ordering
MetricWeekly token order · declared tools supportWindowRolling 7 days · not tool-call count or success rateTemporary official API
  1. #1DDeepSeek: DeepSeek V4 Flash 0423deepseek/deepseek-v4-flashTools declaredCLSSAI live
  2. #2XXiaomi: MiMo-V2.5xiaomi/mimo-v2.5Tools declaredCLSSAI live
  3. #3TTencent: Hy3tencent/hy3Tools declaredCLSSAI live
  4. #4DDeepSeek: DeepSeek V4 Prodeepseek/deepseek-v4-proTools declaredCLSSAI live
  5. #5ZZ.ai: GLM 5.2z-ai/glm-5.2Tools declaredCLSSAI live
  6. #6DDeepSeek: DeepSeek V4 Flash 0731deepseek/deepseek-v4-flash-0731Tools declaredCLSSAI live
  7. #7OOpenAI: GPT-5.6 Lunaopenai/gpt-5.6-lunaTools declaredCLSSAI live
  8. #8NNVIDIA: Nemotron 3 Ultra (free)nvidia/nemotron-3-ultra-550b-a55b:freeTools declaredCLSSAI live
  9. #9MMiniMax: MiniMax M3minimax/minimax-m3Tools declaredCLSSAI live
  10. #10SStepFun: Step 3.7 Flashstepfun/step-3.7-flashTools declaredCLSSAI live

MODALITY

Images

浏览声明图像输出能力的模型，并按公开最近一周模型顺序排列。

Public ordering
MetricImage-output models · weekly token orderWindowRolling 7 days · not generated-image request countTemporary official API
  1. #1GGoogle: Nano Banana (Gemini 2.5 Flash Image)google/gemini-2.5-flash-imageImage output declaredCLSSAI live
  2. #2GGoogle: Nano Banana 2 (Gemini 3.1 Flash Image)google/gemini-3.1-flash-imageImage output declaredCLSSAI live
  3. #3BByteDance Seed: Seedream 4.5bytedance-seed/seedream-4.5Image output declaredOpenRouter preview
  4. #4OOpenAI: GPT Image 2openai/gpt-image-2Image output declaredOpenRouter preview
  5. #5OOpenAI: GPT-5.4 Image 2openai/gpt-5.4-image-2Image output declaredCLSSAI live
  6. #6GGoogle: Nano Banana 2 (Gemini 3.1 Flash Image Preview)google/gemini-3.1-flash-image-previewImage output declaredCLSSAI live
  7. #7XxAI: Grok Imagine Image Qualityx-ai/grok-imagine-image-qualityImage output declaredOpenRouter preview
  8. #8GGoogle: Nano Banana Pro (Gemini 3 Pro Image Preview)google/gemini-3-pro-image-previewImage output declaredCLSSAI live
  9. #9GGoogle: Nano Banana Pro (Gemini 3 Pro Image)google/gemini-3-pro-imageImage output declaredCLSSAI live
  10. #10GGoogle: Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image)google/gemini-3.1-flash-lite-imageImage output declaredCLSSAI live

ECOSYSTEM

Top Apps

按真实模型用量发现应用，并披露排名时间窗口与趋势口径。

Pending

Verified dataset required

Authenticated OpenRouter app-rankings dataset or native CLSSAI app attribution. 此处不使用静态模型、推测份额或合成趋势填充版面。

Track the native implementation
Ranking contract.

名次只来自对应服务端排序结果；页面不会从精选数组、名称关键词或视觉比例推断用量。临时 OpenRouter 数据仅控制展示顺序和元数据，CLSSAI Live 始终只由 api.clssai.com/v1/models 的精确 ID 命中授权。缺失的认证数据、原生聚合、数值指标与趋势会持续登记在 Workbench。