DataLearner 标志

Opus 4.7vsGemini 3.1 Pro Preview

在 12 个共同 benchmark 中，Opus 4.7 整体领先：Opus 4.7 领先 6 项，Gemini 3.1 Pro Preview 领先 5 项，持平 1 项，平均分差 +2.02。

Anthropic
Opus 4.7

Anthropic · 2026-04-16 · 推理大模型

Google Deep Mind
Gemini 3.1 Pro Preview

Google Deep Mind · 2026-02-20 · 多模态大模型

Opus 4.76 项(50%)持平1(42%)5 项Gemini 3.1 Pro Preview

评测分数

按能力类目分组，每组内按分差大小排列；共 12 项。

General Knowledge

Gemini 3.1 Pro Preview 领先 3/5
评测项Opus 4.7Gemini 3.1 Pro Preview分差
HLE54.708 / 157Extended (with tools)51.4015 / 157Thinking High (With Tools)+3.30
ARC-AGI-275.809 / 59最高（无工具）77.107 / 59Thinking High (No Tools)-1.30
MMLU91.506 / 65Normal (No Tools)92.603 / 65Thinking High (No Tools)-1.10
GPQA Diamond94.204 / 178Extended (no tools)94.303 / 178Thinking High (No Tools)-0.10
ARC-AGI-305 / 6Thinking High (No Tools)03 / 6Thinking High (No Tools)持平

AI Agent - Tool Usage

胶着 2/2
评测项Opus 4.7Gemini 3.1 Pro Preview分差
TerminalBench 2.169.708 / 13Thinking High (With Tools)70.707 / 13Thinking High (With Tools)-1
Terminal Bench 2.069.406 / 46Extended (with tools)68.508 / 46Thinking High (With Tools)+0.90

Coding and Software Engineer

Opus 4.7 领先 2/2
评测项Opus 4.7Gemini 3.1 Pro Preview分差
SWE-Bench Pro - Public64.304 / 43Extended (with tools)54.2023 / 43Thinking High (With Tools)+10.10
SWE-bench Verified87.605 / 108Extended (with tools)80.6010 / 108Thinking High (With Tools)+7

Math and Reasoning

Opus 4.7 领先 2/2
评测项Opus 4.7Gemini 3.1 Pro Preview分差
FrontierMath43.806 / 60极高强度思考（无工具）36.9011 / 60Thinking High (No Tools)+6.90
FrontierMath - Tier 422.9012 / 80极高强度思考（无工具）16.7020 / 80Normal (No Tools)+6.20

AI Agent - Information Search

Gemini 3.1 Pro Preview 领先 1/1
评测项Opus 4.7Gemini 3.1 Pro Preview分差
BrowseComp79.3013 / 45Extended (with tools)85.903 / 45Thinking High (With Tools + Internet)-6.60

规格对比

字段Opus 4.7Gemini 3.1 Pro Preview
发布机构AnthropicGoogle Deep Mind
发布时间2026-04-162026-02-20
模型类型推理大模型多模态大模型
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1000K1M
最大输出128K32K

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 4.7Gemini 3.1 Pro Preview
文本输入$5 / 1M tokens$2 / 1M tokens
文本输出$25 / 1M tokens$12 / 1M tokens
缓存读取$0.5 / 1M tokens暂无公开价格
缓存写入$6.25 / 1M tokens暂无公开价格

小结

  • Opus 4.7在以下类目领先:Coding and Software Engineer (2/2)、Math and Reasoning (2/2)
  • Gemini 3.1 Pro Preview在以下类目领先:General Knowledge (3/5)、AI Agent - Information Search (1/1)
  • 胶着类目:AI Agent - Tool Usage

12 个共同 benchmark 上，Opus 4.7 平均高出 2.02 分。

单项差距最大的 benchmark：SWE-Bench Pro - Public — Opus 4.7 64.30，Gemini 3.1 Pro Preview 54.20（分差 +10.10）。

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。

Opus 4.7 详情Gemini 3.1 Pro Preview 详情·在工具里自定义对比