Anthropic
这家公司在各榜上的全部模型，成绩逐榜分开列出。
26 已收录 26 有成绩 5 SOTA
有成绩的模型
- Claude Opus 5.5 SOTA $8.00 57.6
- Claude Sonnet 5.5 SOTA $4.00 56.0
- Claude Fable 5.1 SOTA $20.00 53.4
- Claude Fable 5 SOTA $20.00 49.6
- Claude Haiku 5.5 斩杀者 $0.20 43.4
其余 21 个（不在 SOTA 之列）
- Claude Opus 5 SOTA $10.00 50.8
- Claude Opus 4.8 被斩杀 $10.00 41.8
- Claude Opus 4.7 被斩杀 $10.00 40.7
- Claude Sonnet 5 被斩杀 $4.00 38.2
- Claude Opus 4.6 被斩杀 $10.00 31.9
- Claude Sonnet 4.6 被斩杀 $6.00 30.1
- Claude Opus 4.5 被斩杀 $10.00 29.1
- Claude Opus 4.1 被斩杀 $30.00 22.8
- Claude Sonnet 4.5 被斩杀 $6.00 20.7
- Claude Opus 4 被斩杀 $30.00 20.6
- Claude Sonnet 4 18.9
- Claude Sonnet 3.7 被斩杀 $6.00 17.7
- Claude Haiku 4.5 被斩杀 $2.00 16.9
- Claude Haiku 3.5 8.9
- Claude Opus 3 被斩杀 $30.00 8.7
- Claude Sonnet 3.5 被斩杀 $6.00 7.9
- Claude Sonnet 3 5.9
- Claude 2.1 5.6
- Claude Haiku 3 被斩杀 $0.50 5.6
- Claude 2.0 5.5
- Claude Instant 5.0
在各榜的最好名次
- Arena — Agent Overall #1 Claude Opus 5.5
- Terminal-Bench 4.0 Terminal-Bench 4.0 #1 Claude Opus 5.5
- Arena — Code Overall #1 Claude Opus 5.5
- Arena — Vision Overall (style control) #1 Claude Fable 5
- Arena — Text Overall (style control) #2 Claude Opus 5.5
- Artificial Analysis LLM Leaderboard Artificial Analysis Intelligence Index Claude Opus 5.5
- Artificial Analysis Coding Agents Artificial Analysis Coding Agent Index Claude Sonnet 5.5
- Epoch AI Benchmarking Hub FrontierMath Tiers 1–3 Claude Opus 5.5
- HarnessTax — SWE-bench Lite SWE-bench Lite Claude Fable 5
- HarnessTax — Terminal-Bench 2.0 Terminal-Bench 2.0 Claude Fable 5
旗下模型的性价比位置
圈出的是它在 Artificial Analysis LLM Leaderboard 上有价格也有得分的 4 个当前模型。