Claude Opus
核心事实
时间轴 (近 90 天)
GLM系列模型在Post-Training Bench等榜单上超过了Claude Opus，且价格更低
以Kimi K3实验，给它预填充1%的Claude Opus解码推理片段，其输出风格就会明显向Opus靠拢，某些区段Claude推理链从Kimi K3提取的容易程度比其他模型高出约6个数量级
Databricks在自家一个百万行代码仓库上的基准测试显示，Pi在大部分场景下表现优于Claude Code和Codex，代码质量最高点是Pi + Claude Opus的组合
For Claude Opus, input pricing is $15 per million Tokens and output is $75 per million Tokens
Using top-tier models like Claude Opus or GPT-4 for simple tasks is the most common source of Token waste in Cursor
The Claude Opus series has built a strong reputation in reasoning ability and code generation
Anthropic's Claude Opus is the flagship version of its Claude model family, positioned to handle the most complex reasoning, analysis, and creative tasks
Anthropic、OpenAI等公司的公开定价数据显示每百万Token的推理成本约在0.5-15美元区间
Composer 2.5在Cursor Bench V3.1上得分63.2%，Opus 4.7在默认配置下为61.6%，GPT 5.5为59.2%
全部知识事实 (9)
Composer 2.5在Cursor Bench V3.1上得分63.2%，Opus 4.7在默认配置下为61.6%，GPT 5.5为59.2%
75%已验证Anthropic、OpenAI等公司的公开定价数据显示每百万Token的推理成本约在0.5-15美元区间
65%待验证Anthropic's Claude Opus is the flagship version of its Claude model family, positioned to handle the most complex reasoning, analysis, and creative tasks
100%待验证For Claude Opus, input pricing is $15 per million Tokens and output is $75 per million Tokens
85%待验证The Claude Opus series has built a strong reputation in reasoning ability and code generation
75%待验证Using top-tier models like Claude Opus or GPT-4 for simple tasks is the most common source of Token waste in Cursor
70%待验证GLM系列模型在Post-Training Bench等榜单上超过了Claude Opus，且价格更低
50%待验证以Kimi K3实验，给它预填充1%的Claude Opus解码推理片段，其输出风格就会明显向Opus靠拢，某些区段Claude推理链从Kimi K3提取的容易程度比其他模型高出约6个数量级
50%待验证Databricks在自家一个百万行代码仓库上的基准测试显示，Pi在大部分场景下表现优于Claude Code和Codex，代码质量最高点是Pi + Claude Opus的组合
50%