DataLearner 标志

Claude Fable 5vsGemini 3.1 Pro Preview

在 11 个共同 benchmark 中，Claude Fable 5 整体领先：Claude Fable 5 领先 10 项，Gemini 3.1 Pro Preview 领先 1 项，持平 0 项，平均分差 +84.24。

Anthropic
Claude Fable 5

Anthropic · 2026-06-09 · 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

Claude Fable 510 项(91%)(9%)1 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 11 项。

Coding and Software Engineer

Claude Fable 5 领先 4/4
评测项Claude Fable 5Gemini 3.1 Pro Preview分差
DeepSWE702 / 27Deep Thinking (With Tools)1227 / 27Thinking High (With Tools)+58
SWE-Bench Pro - Public80.301 / 57Deep Thinking (With Tools)54.2034 / 57Thinking High (With Tools)+26.10
WeirdML v287.853 / 52Thinking High (With Tools)72.1017 / 52Normal (With Tools)+15.75
SWE-bench Verified952 / 114Deep Thinking (With Tools)80.6011 / 114Thinking High (With Tools)+14.40

AI Agent - Tool Usage

Claude Fable 5 领先 3/3
评测项Claude Fable 5Gemini 3.1 Pro Preview分差
Terminal-Bench 2.1884 / 44Deep Thinking (With Tools)73.8028 / 44Thinking High (With Tools)+14.20
OSWorld-Verified851 / 26Thinking High (With Tools)76.2012 / 26Thinking (With Tools)+8.80
MCP-Atlas83.305 / 38Normal (With Tools)78.2012 / 38Thinking High (With Tools)+5.10

General Knowledge

胶着 2/2
评测项Claude Fable 5Gemini 3.1 Pro Preview分差
HLE595 / 181Deep Thinking (No Tools)51.4024 / 181Thinking High (With Tools)+7.60
LiveBench78.315 / 115Deep Thinking (No Tools)79.933 / 115Thinking High (No Tools)-1.62

Productivity Knowledge

Claude Fable 5 领先 1/1
评测项Claude Fable 5Gemini 3.1 Pro Preview分差
GDPval-AA v21,7414 / 13Max (With Tools)96512 / 13Thinking (No Tools)+776

常识推理

Claude Fable 5 领先 1/1
评测项Claude Fable 5Gemini 3.1 Pro Preview分差
SimpleBench81.901 / 67Normal (No Tools)79.602 / 67Normal (No Tools)+2.30

规格对比

字段Claude Fable 5Gemini 3.1 Pro Preview
发布机构AnthropicGoogle Deep Mind
发布时间2026-06-092026-02-20
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1M1M
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Claude Fable 5Gemini 3.1 Pro Preview
文本输入$10 / 1M tokens$2 / 1M tokens
文本输出$50 / 1M tokens$12 / 1M tokens
缓存读取$1 / 1M tokens暂无公开价格
缓存写入$12.5 / 1M tokens暂无公开价格

小结

  • Claude Fable 5在以下类目领先:Coding and Software Engineer (4/4)、AI Agent - Tool Usage (3/3)、Productivity Knowledge (1/1)、常识推理 (1/1)
  • 胶着类目:General Knowledge

11 个共同 benchmark 上，Claude Fable 5 平均高出 84.24 分。

单项差距最大的 benchmark：GDPval-AA v2 — Claude Fable 5 1,741，Gemini 3.1 Pro Preview 965（分差 +776）。

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。

Claude Fable 5 详情Gemini 3.1 Pro Preview 详情·在工具里自定义对比