DataLearner 标志

Opus 4.7vsClaude Opus 4.6

在 15 个共同 benchmark 中,Opus 4.7 整体领先:Opus 4.7 领先 11 项,Claude Opus 4.6 领先 2 项,持平 2 项,平均分差 +1.83。

Anthropic
Opus 4.7

Anthropic · 2026-04-16 · 推理大模型

Anthropic
Claude Opus 4.6

Anthropic · 2026-02-05 · 推理大模型

Opus 4.711 (73%)持平2(13%)2 Claude Opus 4.6

评测分数

按能力类目分组,每组内按分差大小排列;共 15 项。

General Knowledge

Opus 4.7 领先 6/7
评测项Opus 4.7Claude Opus 4.6分差
ARC-AGI-275.8011 / 62最高(无工具)66.3017 / 62Extended (no tools)+9.50
GPQA Diamond94.204 / 187Extended (no tools)91.3115 / 187Extended (no tools)+2.89
HLE54.7013 / 172Extended (with tools)5318 / 172Extended (with tools, internet)+1.70
ARC-AGI93.5011 / 68Thinking High (No Tools)9213 / 68Extended (no tools)+1.50
LiveBench76.917 / 115Deep Thinking (No Tools)76.338 / 115Thinking High (No Tools)+0.58
MMLU91.506 / 66Normal (No Tools)91.057 / 66Extended (no tools)+0.45
ARC-AGI-308 / 9Thinking High (No Tools)04 / 9最高(无工具)持平

AI Agent - Tool Usage

Opus 4.7 领先 3/3
评测项Opus 4.7Claude Opus 4.6分差
OSWorld-Verified7810 / 24Extended (with tools)72.7015 / 24Extended (with tools)+5.30
Terminal Bench 2.069.406 / 47Extended (with tools)65.4011 / 47Extended (with tools)+4
MCP-Atlas79.107 / 27Deep Thinking (With Tools)76.8010 / 27Deep Thinking (With Tools)+2.30

Math and Reasoning

Opus 4.7 领先 1/2
评测项Opus 4.7Claude Opus 4.6分差
FrontierMath43.806 / 60极高强度思考(无工具)40.707 / 60最高(无工具)+3.10
FrontierMath - Tier 422.9012 / 80极高强度思考(无工具)22.9012 / 80最高(无工具)持平

AI Agent - Information Search

Claude Opus 4.6 领先 1/1
评测项Opus 4.7Claude Opus 4.6分差
BrowseComp79.3017 / 53Extended (with tools)8411 / 53Thinking (With Tools + Internet)-4.70

Coding and Software Engineer

Opus 4.7 领先 1/1
评测项Opus 4.7Claude Opus 4.6分差
SWE-bench Verified87.606 / 112Extended (with tools)80.8410 / 112Extended (with tools)+6.76

常识推理

Claude Opus 4.6 领先 1/1
评测项Opus 4.7Claude Opus 4.6分差
Simple Bench61.7013 / 63Normal (No Tools)67.608 / 63Normal (No Tools)-5.90

规格对比

字段Opus 4.7Claude Opus 4.6
发布机构AnthropicAnthropic
发布时间2026-04-162026-02-05
模型类型推理大模型推理大模型
架构稠密模型稠密模型
参数规模暂无数据暂无数据
上下文长度1000K1000K
最大输出128K64K

API 调用价格

价格优先使用 DataLearner 配置的 API 记录;缺失项不做推测。

价格项Opus 4.7Claude Opus 4.6
文本输入$5 / 1M tokens$0.5 / 1M tokens
文本输出$25 / 1M tokens$25 / 1M tokens
缓存读取$0.5 / 1M tokens$0.5 / 1M tokens
缓存写入$6.25 / 1M tokens$10 / 1M tokens

小结

  • Opus 4.7在以下类目领先:General Knowledge (6/7)、AI Agent - Tool Usage (3/3)、Math and Reasoning (1/2)、Coding and Software Engineer (1/1)
  • Claude Opus 4.6在以下类目领先:AI Agent - Information Search (1/1)、常识推理 (1/1)

15 个共同 benchmark 上,Opus 4.7 平均高出 1.83 分。

单项差距最大的 benchmark:ARC-AGI-2 — Opus 4.7 75.80,Claude Opus 4.6 66.30(分差 +9.50)。

本页正文由结构化模型、价格与 benchmark 数据生成,不使用实时 LLM 撰写。