SuperCLUE General (March 2026) - Agentic Task Planning: leaderboard
Metric: Score. Source: www.superclueai.com. 24 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.6 (Max) | 89.35 |
| 2 | GPT-5.4 (xHigh) | 80.04 |
| 3 | Gemini 3.1 Pro (Preview) (High) | 78.96 |
| 4 | Kimi K2.5 (Thinking) | 78.44 |
| 5 | DeepSeek V4 Pro (Max) | 77.49 |
| 6 | DeepSeek V4 Flash (Max) | 75.28 |
| 7 | Qwen 3.5 397B A17B (Thinking) | 71.52 |
| 8 | LongCat-Flash-Thinking-2601 | 67.94 |
| 9 | Grok 4.20 Beta (0309) (Reasoning) | 67.35 |
| 10 | GLM-5 | 66.64 |
| 11 | MiniMax-M2.5 | 65.64 |
| 12 | Qwen 3.5 122B A10B (Thinking) | 65.15 |
| 13 | Step 3.5 Flash | 64.06 |
| 14 | Gemini 3 Flash (Preview) (High) | 63.65 |
| 15 | Spark X2 | 62.13 |
Interactive version: theaggregate.ai/benchmark?slug=superclue-general-march-2026-agentic-task-planning · How It Works · Data refreshed daily, snapshot 2026-09-19.