代理式工作
軟體開發以外、取自真實職業的多步任務，比如試算表、文件、客戶服務流程，由模型藉助工具完成。
按它在多少項基準裡進入前 10 排序，一樣多時看最好名次。每列保持各自的排名，不做平均。
| 模型 | 進入前 10 | Artificial Analysis LLM Leaderboard τ³-Banking | Artificial Analysis LLM Leaderboard GDPval-AA | Arena — Agent Overall |
|---|---|---|---|---|
 Claude Fable 5.1 Anthropic | 3/3 | #6 47.2% | #3 62.9% | #1 0.143 |
 Qwen 3.8 Max Alibaba | 2/3 | #1 51.3% | #8 58.6% | #22 0.025 |
| Claude Opus 5.5 Anthropic | 2/3 | — | #1 68.3% | #2 0.138 |
| Claude Sonnet 5.5 Anthropic | 2/3 | — | #2 67.0% | #3 0.125 |
 Muse Spark 1.3 Meta | 2/3 | #3 50.5% | #7 59.2% | #17 0.040 |
 GLM 5.3 Z AI | 2/3 | #4 50.3% | #9 57.6% | #23 0.024 |
| Claude Opus 5 Anthropic | 2/3 | #11 44.7% | #4 61.2% | #7 0.087 |
| GLM 5.3 Flash Z AI | 2/3 | #7 47.2% | #10 57.3% | #31 -0.004 |
 Grok 4.6 SpaceXAI | 1/3 | #2 50.7% | #11 57.3% | #25 0.013 |
 GPT 6 Astra OpenAI | 1/3 | #13 43.1% | #22 52.1% | #4 0.123 |
| Qwen 3.8 Alibaba | 1/3 | #5 48.0% | #35 46.2% | #36 -0.017 |
| Grok 4.7 SpaceXAI | 1/3 | — | #5 60.8% | #15 0.040 |
| GPT 6.1 Sol OpenAI | 1/3 | — | #20 53.8% | #5 0.112 |
 MiMo V2.6 Pro Xiaomi | 1/3 | — | #6 59.3% | #20 0.033 |
| GPT 6 Sol OpenAI | 1/3 | — | #24 50.5% | #6 0.097 |
 Kimi K3 Kimi | 1/3 | #8 46.0% | #23 51.8% | #13 0.042 |
| Claude Fable 5 Anthropic | 1/3 | #22 38.1% | #15 55.6% | #8 0.082 |
 Gemini 3.8 Flash Google | 1/3 | #9 45.8% | #33 46.8% | #21 0.030 |
| Gemini 4 Argon Google | 1/3 | — | #13 56.3% | #9 0.076 |
| Qwen 3.8 Flash Next Alibaba | 1/3 | #10 45.4% | #12 56.6% | #33 -0.007 |
| Claude Opus 4.8 Anthropic | 1/3 | #31 34.2% | #28 47.8% | #10 0.066 |
「—」表示該榜沒有收錄這個模型，不是零分。尚未識別出具體模型的條目不在此處出現，但仍顯示在各自的原始榜單上。