Opus 4.7vsGemini 3.1 Pro Preview
在 16 个共同 benchmark 中，Opus 4.7 整体领先：Opus 4.7 领先 8 项，Gemini 3.1 Pro Preview 领先 7 项，持平 1 项，平均分差 +0.19。
Opus 4.7
Anthropic · 2026-04-16 · 推理大模型
Gemini 3.1 Pro Preview
Google Deep Mind · 2026-02-20 · 多模态大模型
Opus 4.78 项(50%)持平1(44%)7 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 16 项。
General Knowledge
Gemini 3.1 Pro Preview 领先 4/6| 评测项 | Opus 4.7 | Gemini 3.1 Pro Preview | 分差 |
|---|---|---|---|
| HLE | 54.7013 / 172Extended (with tools) | 51.4022 / 172Thinking High (With Tools) | +3.30 |
| LiveBench | 76.917 / 115Deep Thinking (No Tools) | 79.933 / 115Thinking High (No Tools) | -3.02 |
| ARC-AGI-2 | 75.8011 / 62最高（无工具） | 77.109 / 62Thinking High (No Tools) | -1.30 |
| MMLU | 91.506 / 66Normal (No Tools) | 92.603 / 66Thinking High (No Tools) | -1.10 |
| GPQA Diamond | 94.204 / 187Extended (no tools) | 94.303 / 187Thinking High (No Tools) | -0.10 |
| ARC-AGI-3 | 08 / 9Thinking High (No Tools) | 06 / 9Thinking High (No Tools) | 持平 |
AI Agent - Tool Usage
Opus 4.7 领先 3/4| 评测项 | Opus 4.7 | Gemini 3.1 Pro Preview | 分差 |
|---|---|---|---|
| TerminalBench 2.1 | 69.7019 / 27Thinking High (With Tools) | 73.8016 / 27Thinking High (With Tools) | -4.10 |
| OSWorld-Verified | 7810 / 24Extended (with tools) | 76.2011 / 24Thinking (With Tools) | +1.80 |
| MCP-Atlas | 79.107 / 27Deep Thinking (With Tools) | 78.209 / 27Thinking High (With Tools) | +0.90 |
| Terminal Bench 2.0 | 69.406 / 47Extended (with tools) | 68.508 / 47Thinking High (With Tools) | +0.90 |
Coding and Software Engineer
Opus 4.7 领先 2/2| 评测项 | Opus 4.7 | Gemini 3.1 Pro Preview | 分差 |
|---|---|---|---|
| SWE-Bench Pro - Public | 64.307 / 54Extended (with tools) | 54.2032 / 54Thinking High (With Tools) | +10.10 |
| SWE-bench Verified | 87.606 / 112Extended (with tools) | 80.6011 / 112Thinking High (With Tools) | +7 |
Math and Reasoning
Opus 4.7 领先 2/2| 评测项 | Opus 4.7 | Gemini 3.1 Pro Preview | 分差 |
|---|---|---|---|
| FrontierMath | 43.806 / 60极高强度思考（无工具） | 36.9011 / 60Thinking High (No Tools) | +6.90 |
| FrontierMath - Tier 4 | 22.9012 / 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 4.7 | Gemini 3.1 Pro Preview | 分差 |
|---|---|---|---|
| BrowseComp | 79.3017 / 53Extended (with tools) | 85.905 / 53Thinking High (With Tools + Internet) | -6.60 |
常识推理
Gemini 3.1 Pro Preview 领先 1/1| 评测项 | Opus 4.7 | Gemini 3.1 Pro Preview | 分差 |
|---|---|---|---|
| Simple Bench | 61.7013 / 63Normal (No Tools) | 79.602 / 63Normal (No Tools) | -17.90 |
规格对比
| 字段 | Opus 4.7 | Gemini 3.1 Pro Preview |
|---|---|---|
| 发布机构 | Anthropic | Google Deep Mind |
| 发布时间 | 2026-04-16 | 2026-02-20 |
| 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 | 1000K | 1M |
| 最大输出 | 128K | 64K |
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 4.7 | Gemini 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:AI Agent - Tool Usage (3/4)、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 (4/6)、AI Agent - Information Search (1/1)、常识推理 (1/1)
16 个共同 benchmark 上，Opus 4.7 平均高出 0.19 分。
单项差距最大的 benchmark：Simple Bench — Opus 4.7 61.70，Gemini 3.1 Pro Preview 79.60（分差 -17.90）。
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。