各榜单最高分Claude Fable 5.1GPT-6 Astra
开发方AnthropicOpenAI
Artificial Analysis Intelligence Index v4.353.452.7
Terminal-Bench 4.057.958.2
Terminal-Bench 4.0 (AA)55.159.6
Code Arena | WebDev Overall1751.01789.0
Agent Arena · Overall (Ranking / Pareto)14.612.2
Artificial Analysis Coding Agent Index v1.562.261.6
OpenDesign Arena · Quality ranking80.382.7
DeepSWE v1.1—74.1
雷达 IQ(Codex)—110.4
最低真实单价 / 百万 token$0.116$0.052

数据:real-api-pricing数据:codexradar来源与许可