Opus 4.7vsClaude Opus 4.6
在 22 个共同 benchmark 中，Opus 4.7 整体领先：Opus 4.7 领先 18 项，Claude Opus 4.6 领先 4 项，持平 0 项，平均分差 +7.34。
Opus 4.7
Anthropic · 2026-04-16 · 推理大模型
Claude Opus 4.6
Anthropic · 2026-02-05 · 推理大模型
Opus 4.718 项(82%)(18%)4 项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 22 项。
Coding and Software Engineer
Opus 4.7 领先 4/4| 评测项 | Opus 4.7 | Claude Opus 4.6 | 分差 |
|---|---|---|---|
| GSO | 44.121 / 21Normal (With Tools) | 33.335 / 21Normal (With Tools) | +10.79 |
| WeirdML v2 | 76.4013 / 52Normal (With Tools) | 65.9023 / 52Normal (With Tools) | +10.50 |
| Text Arena (Coding) | 1,5627 / 35Normal (No Tools) | 1,5558 / 35Normal (No Tools) | +7.04 |
| SWE-bench Verified | 87.606 / 116Extended (with tools) | 80.8410 / 116Extended (with tools) | +6.76 |
Agent Level Benchmark
Opus 4.7 领先 2/3| 评测项 | Opus 4.7 | Claude Opus 4.6 | 分差 |
|---|---|---|---|
| τ³-Banking | 40.2130 / 164Max (With Tools) | 27.3269 / 164Max (With Tools) | +12.89 |
| τ²-Bench - Telecom | 74128 / 264Normal (With Tools) | 84.8090 / 264Normal (With Tools) | -10.80 |
| Terminal Bench Hard | 54.5015 / 244Normal (With Tools) | 48.5025 / 244Normal (With Tools) | +6 |
AI Agent - Tool Usage
Opus 4.7 领先 3/3| 评测项 | Opus 4.7 | Claude Opus 4.6 | 分差 |
|---|---|---|---|
| OSWorld-Verified | 7811 / 26Extended (with tools) | 72.7016 / 26Extended (with tools) | +5.30 |
| Terminal Bench 2.0 | 69.406 / 48Extended (with tools) | 65.4011 / 48Extended (with tools) | +4 |
| MCP-Atlas | 79.1013 / 43Deep Thinking (With Tools) | 76.8018 / 43Deep Thinking (With Tools) | +2.30 |
General Knowledge
Opus 4.7 领先 2/2| 评测项 | Opus 4.7 | Claude Opus 4.6 | 分差 |
|---|---|---|---|
| HLE | 33.30184 / 563Normal (No Tools) · Text only | 19.10299 / 563Normal (No Tools) · Text only | +14.20 |
| CritPt | 5.1094 / 200Normal (No Tools) | 2.80120 / 200Normal (No Tools) | +2.30 |
Math and Reasoning
Opus 4.7 领先 2/2| 评测项 | Opus 4.7 | Claude Opus 4.6 | 分差 |
|---|---|---|---|
| FrontierMath Tier 4 v2 | 31.7117 / 42Max (No Tools) | 26.8323 / 42Max (No Tools) | +4.88 |
| FrontierMath v2 | 70.1815 / 58Max (No Tools) | 65.9618 / 58Max (No Tools) | +4.21 |
Claw-style Agent Evaluation
Opus 4.7 领先 1/1| 评测项 | Opus 4.7 | Claude Opus 4.6 | 分差 |
|---|---|---|---|
| PinchBench v2 | 76.0115 / 45Reported best (effort unspecified) | 69.9026 / 45Reported best (effort unspecified) | +6.11 |
Instruction Following
Claude Opus 4.6 领先 1/1| 评测项 | Opus 4.7 | Claude Opus 4.6 | 分差 |
|---|---|---|---|
| IF Bench | 43.60189 / 282Normal (No Tools) | 44.60182 / 282Normal (No Tools) | -1 |
Long Context
Opus 4.7 领先 1/1| 评测项 | Opus 4.7 | Claude Opus 4.6 | 分差 |
|---|---|---|---|
| AA-LCR | 75.7082 / 170Normal (No Tools) | 67116 / 170Normal (No Tools) | +8.70 |
Multimodal Understanding
Opus 4.7 领先 1/1| 评测项 | Opus 4.7 | Claude Opus 4.6 | 分差 |
|---|---|---|---|
| MMMU-Pro | 76.4073 / 227Normal (No Tools) | 72.50114 / 227Normal (No Tools) | +3.90 |
Text Embedding
Claude Opus 4.6 领先 1/1| 评测项 | Opus 4.7 | Claude Opus 4.6 | 分差 |
|---|---|---|---|
| Context Arena | 22.12122 / 126Normal (No Tools) | 60.3376 / 126Normal (No Tools) | -38.21 |
Writing and Creative Capabilities
Opus 4.7 领先 1/1| 评测项 | Opus 4.7 | Claude Opus 4.6 | 分差 |
|---|---|---|---|
| Creative Writing | 1,9079 / 106Normal (No Tools) | 1,80419 / 106Normal (No Tools) | +103 |
常识推理
Claude Opus 4.6 领先 1/1| 评测项 | Opus 4.7 | Claude Opus 4.6 | 分差 |
|---|---|---|---|
| SimpleBench | 61.7027 / 93Normal (No Tools) | 67.6019 / 93Normal (No Tools) | -5.90 |
科学与综合推理
Opus 4.7 领先 1/1| 评测项 | Opus 4.7 | Claude Opus 4.6 | 分差 |
|---|---|---|---|
| GPQA Diamond | 88.50110 / 462Normal (No Tools) | 84176 / 462Normal (No Tools) | +4.50 |
规格对比
| 字段 | Opus 4.7 | Claude Opus 4.6 |
|---|---|---|
| 发布机构 | Anthropic | Anthropic |
| 发布时间 | 2026-04-16 | 2026-02-05 |
| 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 | 1000K | 1000K |
| 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 | Claude Opus 4.6 |
|---|---|---|
| 文本输入 | $5 / 1M tokens | $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 | $6.25 / 1M tokens |
小结
- Opus 4.7在以下类目领先:Coding and Software Engineer (4/4)、Agent Level Benchmark (2/3)、AI Agent - Tool Usage (3/3)、General Knowledge (2/2)、Math and Reasoning (2/2)、Claw-style Agent Evaluation (1/1)、Long Context (1/1)、Multimodal Understanding (1/1)、Writing and Creative Capabilities (1/1)、科学与综合推理 (1/1)
- Claude Opus 4.6在以下类目领先:Instruction Following (1/1)、Text Embedding (1/1)、常识推理 (1/1)
22 个共同 benchmark 上，Opus 4.7 平均高出 7.34 分。
单项差距最大的 benchmark：Creative Writing — Opus 4.7 1,907，Claude Opus 4.6 1,804（分差 +103）。
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。