独立 AI 分析
了解 AI 格局，为你的使用场景选择最佳模型和服务商
新增语言模型评测 · 11月2日
Gemini 4 Carbon (Low)新增语言模型评测 · 11月2日
Nanbeige4.3-33B新增语言模型评测 · 11月1日
Gemini 4 Carbon (Medium)新增语言模型评测 · 11月1日
DeepSeek V4.1 Flash 1017 (High)新增语言模型评测 · 10月30日
DeepSeek V4.1 Flash 1017 (Non-reasoning)新增语言模型评测 · 10月29日
Mistral Large 4 (Medium)新增语言模型评测 · 10月29日
Mistral Large 4 (High)新文章发布 · 10月28日
Gemini 4 Carbon is the new #1 on the Artificial Analysis Intelligence Index v5
新文章发布 · 10月28日
GitHub Copilot Fusion orchestrates Anthropic and OpenAI models to balance performance and cost
新增语言模型评测 · 10月28日
Gemini 4 Carbon (High)新文章发布 · 10月27日
Announcing Artificial Analysis Intelligence Index v5
方法论更新 · 10月27日
Artificial Analysis Intelligence Index v5.0新文章发布 · 10月26日
GPT-6.1 Terra pushes the cost efficiency frontier
新文章发布 · 10月26日
Code2Video-AA: how well do AI models design videos in code?
新增语言模型评测 · 10月26日
GPT-6.1 Terra (Non-reasoning)新增语言模型评测 · 10月26日
GPT-6.1 Terra (Low)新增语言模型评测 · 10月26日
GPT-6.1 Terra (Medium)新增语言模型评测 · 10月26日
GPT-6.1 Terra (High)新增语言模型评测 · 10月26日
GPT-6.1 Terra (Xhigh)新增语言模型评测 · 10月26日
GPT-6.1 Terra (Max)查看更多Artificial Analysis Intelligence Index
Artificial Analysis Intelligence Index v5.0 incorporates 9 evaluations: AA-Briefcase v1.1, AutomationBench-AA, Terminal-Bench 4.0, FrontierCode 2.0, AA-Omniscience v1.1, GDP.pdf, HLE-Diamond, CritPt v2 (Hard Subset), and Terminal-Bench-Science 0.1
No publicly available API
每项 Intelligence Index 任务的成本
Weighted average cost (USD) per Intelligence Index task · Segmented by token type · Lower is better
No publicly available API
Intelligence Index 与每项任务的成本
Artificial Analysis Intelligence Index · Weighted average cost (USD) per Artificial Analysis Intelligence Index task
Most attractive quadrant
Pareto line
按模型发布版本划分的 Intelligence Index 与单任务成本
All reasoning and effort variants of each selected release · Weighted average cost (USD) per Artificial Analysis Intelligence Index task
Most attractive quadrant
Pareto line
Artificial Analysis Progress Index v1 变化趋势
Item response theory estimate of general capability across Artificial Analysis evaluations · Highest-scoring effort variant shown
Artificial Analysis Coding Agent Index
Artificial Analysis Coding Agent Index v2.0 incorporates 3 benchmarks: FrontierCode v1.1, Terminal-Bench 4.0, and SWE-Atlas-QnA-AA · Higher is better
着色方式
未公开
Artificial Analysis Cyber Index
Artificial Analysis Cyber Index v1 incorporates 3 evaluations: CWE-Bench-AA, DeepsecBench-AA, and CyberGym-E2E-AA
Successes
Safety blocks
Trusted-access model (not publicly available)
Artificial Analysis Decision Index
Weighted average of AA-Omniscience-Decision, HLE-MultiChoice, When2Call, and WildGuardTest · Higher is better
能力指数对比
所选模型在各能力指数中的排名
| 财务与会计 | 战略与运营 | 法律 | 医疗与健康 | 工程 | 经济学 | |
|---|---|---|---|---|---|---|
| Claude Opus 5.5 (Max, Default Fallback) | 1 (59) | 1 (61) | 1 (58) | 1 (58) | 1 (65) | 1 (62) |
| GPT-6 Astra (Max) | 2 (57) | 3 (58) | 2 (56) | 3 (54) | 2 (62) | 3 (58) |
| GPT-6.1 Sol (Max) | 3 (56) | 7 (58) | 4 (54) | 4 (52) | 3 (59) | 4 (57) |
| Claude Sonnet 5.5 (Max, Default Fallback) | 4 (56) | 2 (58) | 5 (52) | 2 (54) | 4 (58) | 5 (56) |
| Claude Fable 5.1 (Max, Default Fallback) | 6 (54) | 4 (58) | 3 (54) | 7 (51) | 5 (57) | 2 (59) |
| Muse Wonder Full (Max) | 5 (55) | 5 (58) | 6 (51) | 5 (52) | 6 (55) | 6 (56) |
| Kimi K3.1 Pro | 7 (53) | 6 (58) | 7 (49) | 6 (52) | 8 (50) | 7 (51) |
| Muse Wonder (Max) | 9 (51) | 9 (54) | 9 (47) | 8 (49) | 9 (49) | 8 (51) |
| GPT-6.1 Terra (Max) | 10 (49) | 10 (50) | 10 (46) | 10 (45) | 7 (54) | 9 (49) |
| Grok 4.7 (Xhigh) | 8 (52) | 8 (55) | 8 (47) | 9 (47) | 11 (42) | 10 (48) |
| Qwen3.8 Max (0902) | 11 (45) | 13 (47) | 12 (41) | 12 (41) | 12 (39) | 12 (44) |
| DeepSeek V4.1 Flash 1017 (Max) | 14 (42) | 11 (49) | 14 (38) | 11 (41) | 15 (37) | 14 (40) |
| Gemini 3.8 Flash (High) | 13 (42) | 14 (47) | 11 (42) | 14 (38) | 13 (38) | 13 (41) |
| Claude Haiku 5.5 (Max) | 12 (44) | 15 (39) | 13 (39) | 15 (37) | 10 (42) | 11 (44) |
| GLM-5.3 (Max) | 15 (42) | 12 (47) | 15 (36) | 13 (39) | 14 (38) | 15 (39) |
Evaluations
Evaluations measured independently by Artificial Analysis · AA-PersonalityBench scores are percentile ranks from 1 to 100
No publicly available API
Agentic knowledge work, Elo
Knowledge
1 - hallucination rate
Agentic SaaS workflows
Agentic scientific research workflows in a terminal
Agentic software engineering in open-source repositories
Reasoning & knowledge
Professional document reasoning, All-pass
Agentic coding & terminal use
Physics reasoning
Quantitative analysis on spreadsheets & documents
Legal agentic work, Hallucination-Gated All-Pass Rate
Agentic business operations
Agentic real-world work tasks, Elo
Agentic interactive 3D experiences, Elo
Product and brand videos designed in code, Elo
Kubernetes incident root-cause analysis
Medical long context reasoning
Visual reasoning
Rich, varied, well-crafted language
Well-structured, scannable, answer-first
Truthful, acknowledges uncertainty, non-sycophantic
Use of filler and generic language