AA APEX-Agents: leaderboard
Artificial Analysis APEX-Agents: 452 long-horizon professional tasks across investment banking, management consulting, and corporate law testing agentic capabilities.
Metric: Pass@1 (%). Source: artificialanalysis.ai. Saturation forecast: Around June 2027. 33 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.5 Flash (High) | 47.05 | #60 (Gemini 3.5 Flash) |
| 2 | Qwen 3.8 Max (0902) | 42.4 | #29 |
| 3 | Kimi K3 (Max) | 41.3 | #30 (Kimi K3) |
| 4 | GPT-5.6 Terra (Max) | 38.94 | #41 (GPT-5.6 Terra) |
| 5 | Step 5 Preview | 37.98 | #44 |
| 6 | GPT-5.5 (xHigh) | 37.68 | #31 (GPT-5.5) |
| 7 | GPT-5.6 Luna (Max) | 35.84 | #68 (GPT-5.6 Luna) |
| 8 | GLM-5.2 (Max) | 33.7 | #73 (GLM-5.2) |
| 9 | GPT-5.4 (xHigh) | 33.26 | #79 (GPT-5.4) |
| 10 | Claude Opus 4.6 (Max) | 33.04 | #63 (Claude Opus 4.6) |
| 11 | Gemini 3.1 Pro (Preview) | 32.01 | #58 |
| 12 | Apodex 1.1 | 31.19 | #87 |
| 13 | Kimi K2.6 (Thinking) | 28.47 | #104 (Kimi K2.6) |
| 14 | GPT-5.4 Mini (xHigh) | 28.17 | #212 (GPT-5.4 Mini) |
| 15 | Claude Sonnet 4.6 (Max) | 28.02 | #90 (Claude Sonnet 4.6) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=aa-apex-agents · How It Works · Data refreshed daily, snapshot 2026-10-20.