SuperCLUE-Agent (October 2023) - Long & Short-term Memory: leaderboard
SuperCLUE-Agent evaluates Chinese LLM agent capabilities: task planning, tool use and long- and short-term memory. October 2023 edition (2023年10月): memory total covering multi-document QA, long-range dialogue and few-shot example learning. Each dated workbook is a separate evaluation round, kept apart from the current board. Score 0-100, higher is better.
Metric: Score. Source: www.superclueai.com. Saturation forecast: Estimated already saturated. 19 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-4 | 66.67 | #459 |
| 2 | GPT-3.5 Turbo | 55.74 | #858 |
| 3 | Baichuan2-13B-Chat | 52.74 | #1210 |
| 4 | Chinese-Alpaca-2-13B | 42.62 | #1390 |
| 5 | MiniMax-abab5.5 | 42.23 | #1030 |
| 6 | Qwen-7B Chat | 37.67 | #1353 |
| 7 | Llama 2 13B Chat | 34.35 | #1414 |
| 8 | ChatGLM2 6B | 32.33 | #1475 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=superclue-agent-october-2023-long-short-term-memory · How It Works · Data refreshed daily, snapshot 2026-11-05.