开放权重模型比较
从质量、性能、推理速度、上下文窗口、参数量和许可证详情等关键指标比较和分析开放权重 AI 模型。
如果模型权重可供下载，我们便将其视为开放权重模型。这样用户就可以在自己的基础设施上自行托管，并通过微调等方式定制模型。
如需了解我们方法论的更多详情，请参阅常见问题。
GLM-5.3 (max) 和 
Kimi K3 (max) 是智能得分最高的开放权重模型，其次是 GLM-5.3-Flash 和 
Qwen3.8 2.4T A95B。亮点
开放性
Artificial Analysis Openness Index: Score
Openness Index assesses model openness on a 0 to 100 normalized scale (higher is more open)
开放权重模型进展
Progress in Open Weights vs. Proprietary Intelligence
Artificial Analysis Intelligence Index v4.3 incorporates 10 evaluations: AA-Briefcase, GDPval-AA v2, AutomationBench-AA, Terminal-Bench v4.0, SciCode, Humanity's Last Exam, GDP.pdf, CritPt, AA-Omniscience, AA-LCR v1.1
各实验室开放权重语言模型智能随时间的变化
各规模开放权重模型智能随时间的变化
Artificial Analysis Intelligence Index v4.3 incorporates 10 evaluations: AA-Briefcase, GDPval-AA v2, AutomationBench-AA, Terminal-Bench v4.0, SciCode, Humanity's Last Exam, GDP.pdf, CritPt, AA-Omniscience, AA-LCR v1.1
智能
Artificial Analysis Intelligence Index
Artificial Analysis Intelligence Index v4.3 incorporates 10 evaluations: AA-Briefcase, GDPval-AA v2, AutomationBench-AA, Terminal-Bench v4.0, SciCode, Humanity's Last Exam, GDP.pdf, CritPt, AA-Omniscience, AA-LCR v1.1
Intelligence Evaluations
Intelligence evaluations measured independently by Artificial Analysis · Higher is better
Agentic knowledge work, (Elo-500)/2000
Agentic real-world work tasks, (Elo-500)/2000
AutomationBench-AAUpdated
Agentic SaaS workflows
Agentic coding & terminal use
Coding
Reasoning & knowledge
GDP.pdfNew
Professional document reasoning, All-pass
Physics reasoning
Knowledge
1 - hallucination rate
Long context reasoning
Legal agentic work, criterion pass rate
Agentic business operations
Quantitative analysis on spreadsheets & documents
Instruction following
Agentic tool use
Long-horizon agentic tasks
Kubernetes incident root-cause analysis
Visual reasoning
规模
按模型规模划分的 Intelligence Index
Artificial Analysis Intelligence Index v4.3 incorporates 10 evaluations: AA-Briefcase, GDPval-AA v2, AutomationBench-AA, Terminal-Bench v4.0, SciCode, Humanity's Last Exam, GDP.pdf, CritPt, AA-Omniscience, AA-LCR v1.1
Model Size: Total and Active Parameters
Comparison between total model parameters and parameters active during inference
Intelligence Index vs. Active Parameters
Artificial Analysis Intelligence Index · Active parameters at inference time
Most attractive quadrant
Pareto line
Intelligence Index vs. Total Parameters
Artificial Analysis Intelligence Index · Size in parameters (billions)
Most attractive quadrant
Pareto line
上下文窗口
Context Window
Context window: tokens limit · Higher is better
更多详情
权重 | 服务商基准测试 | ||||||||
|---|---|---|---|---|---|---|---|---|---|
GLM-5.3 (max) | Z AI | 45 | 753B 推理时启用 40B 个参数 | 1M | $0.9 | 72 |   +14 | ||
Kimi K3 (max) | Kimi | 44 | 2.8T 推理时启用 104B 个参数 | 1M | $2.3 | 36 |    +12 | ||
GLM-5.3-Flash | Z AI | 42 | 320B 推理时启用 18B 个参数 | 1M | $0.1 | 107 |   +16 | ||
Qwen3.8 2.4T A95B | Alibaba | 40 | 2.4T 推理时启用 95B 个参数 | 984k | $1.2 | 40 |  +3 | ||
DeepSeek V4.1 Flash (Reasoning, Max Effort) |  DeepSeek | 40 | 552B 推理时启用 16B 个参数 | 1M | $0.2 | 214 | | ||
DeepSeek V4 Pro 0813 (Reasoning, Max Effort) | DeepSeek | 36 | 1.6T 推理时启用 49B 个参数 | 1M | $0.7 | 82 |   +7 | ||
Qwen3.8 27B (xhigh) | Alibaba | 34 | 27B | 256k | $0.4 | 45 |  +5 | ||
K2 Horizon 375B A23B |  MBZUAI Institute of Foundation Models | 31 | 375B 推理时启用 23B 个参数 | 524k | - | - | - | ||
MiniMax-M3 |  MiniMax | 30 | 428B 推理时启用 23B 个参数 | 1M | $0.2 | 103 |  +12 | ||
Inkling (xhigh) |  Thinking Machines | 26 | 975B 推理时启用 41B 个参数 | 1M | $0.7 | 91 | +3 | ||
Nemotron 3 Ultra 550B A55B (Reasoning) |  NVIDIA | 23 | 550B 推理时启用 55B 个参数 | 262k | $0.5 | 206 | +6 | ||
Muse Glimmer (high) |  Meta | 18 | 30B | 131k | $0.2 | 90 | | ||
Mistral Medium 3.5 |  Mistral | 15 | 128B | 256k | $1.2 | 148 | | ||
gpt-oss-120b (high) |  OpenAI | 12 | 117B 推理时启用 5.1B 个参数 | 131k | $0.2 | 228 |    +17 |