模型排行榜
AIHOT评分为综合LMArena、Artificial Analysis、LLM Stats、Vellum、LiveBench等7大权威榜单的加权得分。成本为API标准定价(USD/百万token)。数据更新:2026-08-17。
| 模型名称 |
上线日期 |
输入成本 |
输出成本 |
AIHOT评分 |
| Claude Opus 5 |
2026-07-24 |
$5.00 |
$25.00 |
89.7 |
| Muse Spark 1.2 |
2026-08-01 |
— |
— |
86.7 |
| Claude Fable 5 |
2026-06-09 |
$10.00 |
$50.00 |
80.3 |
| GPT-5.6 Terra |
2026-07-09 |
$2.50 |
$15.00 |
72.2 |
| Kimi K3 |
2026-07-16 |
$3.00 |
$15.00 |
71.4 |
| GPT-5.6 Sol |
2026-07-09 |
$5.00 |
$30.00 |
70.0 |
| DeepSeek V4 Flash |
2026-07 |
$0.14 |
$0.28 |
66.7 |
| Claude Sonnet 5 |
2026-07 |
$2.00 |
$10.00 |
58.3 |
| Grok 4.6 |
2026-08-13 |
$2.00 |
$6.00 |
新† |
| Grok 4.5 |
2026-06 |
$2.00 |
$6.00 |
55.0 |
| Claude Opus 4.8 |
2026-04 |
$5.00 |
$25.00 |
53.0 |
| Gemini 3.1 Pro |
2026-02 |
$2.00 |
$12.00 |
48.8 |
| Gemini 3.6 Flash |
2026-07 |
$1.50 |
$7.50 |
43.8 |
| GLM 5.2 |
2026-06 |
— |
— |
41.0 |
| GPT-5.6 Luna |
2026-07-09 |
$1.00 |
$6.00 |
32.8 |
| Gemini 3.5 Flash |
2026-05 |
$1.50 |
$9.00 |
31.7 |
† Grok 4.6为新发布模型,AA Briefcase榜排名第2(Elo 1577),综合分待统计。
输入/输出成本单位为美元(USD)/百万token,短上下文基础费率。
数据来源:AI Rankings · AI Cost Check · Frontier Model Comparison