Claude Mythos PreviewvsClaude Opus 4.6

在 7 个共同 benchmark 中，Claude Mythos Preview 整体领先：Claude Mythos Preview 领先 7 项，Claude Opus 4.6 领先 0 项，持平 0 项，平均分差 +9.68。

Anthropic · 2026-04-07 · 聊天大模型

Anthropic · 2026-02-05 · 推理大模型

Claude Mythos Preview7 项(100%)(0%)0 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 7 项。

Claude Mythos Preview 领先 2/2

评测项	Claude Mythos Preview	Claude Opus 4.6	分差
Terminal Bench 2.0	822 / 46Extended (with tools)	65.4011 / 46Extended (with tools)	+16.60
OSWorld-Verified	79.603 / 18Extended (with tools)	72.709 / 18Extended (with tools)	+6.90

Claude Mythos Preview 领先 2/2

评测项	Claude Mythos Preview	Claude Opus 4.6	分差
SWE-bench Multilingual	87.301 / 20Extended (with tools)	7212 / 20Extended (with tools)	+15.30
SWE-bench Verified	93.903 / 108Extended (with tools)	80.849 / 108Extended (with tools)	+13.06

Claude Mythos Preview 领先 2/2

评测项	Claude Mythos Preview	Claude Opus 4.6	分差
HLE	64.701 / 157Extended (with tools)	5311 / 157Extended (with tools, internet)	+11.70
GPQA Diamond	94.601 / 178Extended (no tools)	91.3114 / 178Extended (no tools)	+3.29

Claude Mythos Preview 领先 1/1

评测项	Claude Mythos Preview	Claude Opus 4.6	分差
BrowseComp	84.904 / 45Extended (with tools)	847 / 45Thinking (With Tools + Internet)	+0.90

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

Claude Mythos Preview在以下类目领先:AI Agent - Tool Usage (2/2)、Coding and Software Engineer (2/2)、General Knowledge (2/2)、AI Agent - Information Search (1/1)

7 个共同 benchmark 上，Claude Mythos Preview 平均高出 9.68 分。

单项差距最大的 benchmark：Terminal Bench 2.0 — Claude Mythos Preview 82，Claude Opus 4.6 65.40（分差 +16.60）。

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。