FutureX - Level 4 (Super Agent): leaderboard
FutureX Level 4 (Super Agent tier): open-ended questions about highly volatile future events that demand wide search and reasoning under uncertainty. Official runs from the newest week that has them, so the event set changes weekly. Higher is better, 0-100 (%); weighted 40% in the FutureX overall score.
Metric: Level 4 score (latest official week, %). Source: futurex.live. Saturation forecast: Around 2029. 14 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.4 (High) | 44.38 | #83 (GPT-5.4) |
| 2 | Seed 2.0 Pro | 40.22 | #122 |
| 3 | DeepSeek V3.2 (Thinking) | 33.79 | #210 (DeepSeek V3.2) |
| 4 | Qwen 3.7 Max | 33.51 | #85 |
| 5 | GPT-5.5 (High) | 33.42 | #33 (GPT-5.5) |
| 6 | Grok 4.20 (Reasoning) | 29.59 | #179 (Grok 4.20) |
| 7 | Gemini 3.1 Pro (Preview) (High) | 29.04 | #64 (Gemini 3.1 Pro (Preview)) |
| 8 | Seed 2.0 Lite | 27.25 | #161 |
| 9 | MiniMax-M3 | 26.92 | #151 |
| 10 | MiniMax-M2.7 | 26.21 | #270 |
| 11 | Gemini 3.5 Flash | 21.74 | #66 |
| 12 | Qwen 3.6 Max Preview | 18.7 | #91 |
| 13 | Grok 4 | 10.98 | #189 |
| 14 | DeepSeek V4 Flash | 7.54 | #155 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=futurex-level-4-super-agent · How It Works · Data refreshed daily, snapshot 2026-11-05.