AIHOT LEADERBOARD
AI 模型排行榜
职业交付件、业务流程、金融、法律、投行咨询长任务、客服、数据分析与医疗编码，看专业工作能不能交付。
10 项评测 · 4 家机构 · 10/10 08:06 更新
数据同步自 AIHOT 官方模型榜 ↗，沿用官方排名与评分。
专业办公榜 30 个模型
| 名次 | 模型 | |||
|---|---|---|---|---|
| 01 | Gemini 4 ArgonGoogle | 2026-09-30 | 7 项评测覆盖 75% | 76.4 |
| 02 | Claude Opus 5.5Anthropic | 2026-09-22 | 10 项评测覆盖 100% | 71.3 |
| 03 | Claude Sonnet 5.5Anthropic | 2026-09-28 | 10 项评测覆盖 100% | 70.5 |
| 04 | Claude Fable 5.1Anthropic | 2026-09-01 | 10 项评测覆盖 100% | 70.1 |
| 05 | Claude Opus 5Anthropic | 2026-07-24 | 10 项评测覆盖 100% | 68.9 |
| 06 | Claude Fable 5Anthropic | 2026-06-09 | 9 项评测覆盖 88% | 68.7 |
| 07 | Muse Spark 1.3Meta | 2026-09-02 | 8 项评测覆盖 88% | 68.3 |
| 08 | GPT-6 AstraOpenAI | 2026-09-03 | 10 项评测覆盖 100% | 67.2 |
| 09 | Gemini 3.7 FlashGoogle | 2026-08-13 | 9 项评测覆盖 88% | 65.9 |
| 10 | Grok 4.6xAI | 2026-08-12 | 10 项评测覆盖 100% | 65.8 |
| 11 | GPT-6.1 SolOpenAI | 2026-09-29 | 10 项评测覆盖 100% | 65.8 |
| 12 | Gemini 3.8 FlashGoogle | 2026-09-02 | 9 项评测覆盖 100% | 65.2 |
| 13 | MiMo-V2.6-ProXiaomi | 2026-09-21 | 7 项评测覆盖 75% | 65.2 |
| 14 | Grok 4.7xAI | 2026-09-21 | 9 项评测覆盖 100% | 64.4 |
| 15 | Kimi K3Moonshot AI开源权重 ↗ | 2026-07-16 | 10 项评测覆盖 100% | 63.8 |
| 16 | Qwen3.8 MaxAlibaba | 2026-07-19 | 9 项评测覆盖 100% | 63.5 |
| 17 | GPT-6 SolOpenAI | 2026-09-22 | 9 项评测覆盖 100% | 62.9 |
| 18 | MiMo V 2.6 FlashXiaomi | 2026-09-21 | 6 项评测覆盖 63% | 62.8 |
| 19 | GLM-5.3Z.ai开源权重 ↗ | 2026-08-18 | 9 项评测覆盖 100% | 62.7 |
| 20 | GPT-5.6 SolOpenAI | 2026-07-09 | 8 项评测覆盖 75% | 62.7 |
| 21 | GLM-5.3 FlashZ.ai开源权重 ↗ | 2026-08-26 | 8 项评测覆盖 88% | 62.7 |
| 22 | Claude Opus 4.7Anthropic | 2026-04-16 | 6 项评测覆盖 63% | 62.6 |
| 23 | Claude Opus 4.8Anthropic | 2026-05-28 | 9 项评测覆盖 88% | 62.4 |
| 24 | Qwen3.8 Flash NextAlibaba | 2026-08-26 | 4 项评测覆盖 38% | 62.2 |
| 25 | Muse Spark 1.2Meta | 2026-08-05 | 8 项评测覆盖 88% | 62.1 |
| 26 | Step 5 PreviewStepFun | 2026-09-18 | 6 项评测覆盖 63% | 61.8 |
| 27 | GPT-5.6 TerraOpenAI | 2026-07-09 | 9 项评测覆盖 100% | 61.6 |
| 28 | DeepSeek V4.1 FlashDeepSeek开源权重 ↗ | 2026-09-10 | 9 项评测覆盖 100% | 61.1 |
| 29 | GPT-5.5OpenAI | 2026-04-23 | 9 项评测覆盖 88% | 60.9 |
| 30 | Grok 4.5xAI | 2026-07-08 | 9 项评测覆盖 88% | 60.7 |
AIHOT 评分把各家评测换算到同一条能力刻度上：50 分是参照组模型的平均水平，每 15 分约是一个标准差。评分不是正确率。分类评分是模型在这个领域的能力，只列这个领域测得够多的模型。 价格为 API 参考报价，原始出处见模型详情。