APEX-Agents: leaderboard
The AI Productivity Index for Agents (APEX-Agents) measures whether frontier AI agents can execute long-horizon, cross-application tasks across three jobs in professional services.
Metric: Mean Score (ReAct) (imported). Source: mercor.com. Source: benchmarklist.com. Saturation forecast: Around September 2027. 45 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.5 Flash | 66.1 | #60 |
| 2 | GPT-5.4 (xHigh) | 52.7 | #79 (GPT-5.4) |
| 3 | Claude Opus 4.7 (Max) | 50.6 | #49 (Claude Opus 4.7) |
| 4 | Claude Opus 4.6 (Max) | 48.4 | #63 (Claude Opus 4.6) |
| 5 | GPT-5.2 (xHigh) | 48.4 | #110 (GPT-5.2) |
| 6 | Gemini 3.1 Pro (Preview) (High) | 48.2 | #58 (Gemini 3.1 Pro (Preview)) |
| 7 | GPT-5.3 Codex (High) | 46.9 | #69 (GPT-5.3 Codex) |
| 8 | Claude Opus 4.6 (High) | 45.6 | #63 (Claude Opus 4.6) |
| 9 | GPT-5.2 Codex (High) | 42.2 | #95 (GPT-5.2 Codex) |
| 10 | Kimi K3 (Max) | 41 | #30 (Kimi K3) |
| 11 | Claude Sonnet 4.6 (High) | 40.7 | #90 (Claude Sonnet 4.6) |
| 12 | GPT-5.6 Sol (Max) | 39.9 | #19 (GPT-5.6 Sol) |
| 13 | Gemini 3 Flash (High) | 39.5 | #99 (Gemini 3 Flash) |
| 14 | Claude Opus 4.8 (Max) | 39.4 | #37 (Claude Opus 4.8) |
| 15 | GPT-5.5 (xHigh) | 38.5 | #31 (GPT-5.5) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=apex-agents · How It Works · Data refreshed daily, snapshot 2026-10-20.