ImplicitMemBench: leaderboard
ImplicitMemBench (2026) measures implicit memory in LLMs, whether behavior adapts without explicit recall after learning, interference and test phases in a conversation. This board is the overall score, the mean of procedural-memory accuracy, classical-conditioning accuracy and the judged priming score; higher is better.
Metric: Overall score: mean of procedural-memory accuracy, classical-conditioning accuracy (%) and the GPT-4o-mini-judged priming influence score (0 to 95 bands) on the 300 ImplicitMemBench items (learning, interference and test phases, first-attempt scoring, mean of three runs at temperature 0); higher is better. Source: arxiv.org. Saturation forecast: Around August 2028. 17 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | DeepSeek R1 | 65.3 | #482 | |
| 2 | Qwen 3 32B | 64.13 | #972 | |
| 3 | GPT-5 | 63 | #178 | |
| 4 | Qwen 3 8B | 62.35 | #1873 | |
| 5 | O3 | 61.79 | #237 | |
| 6 | O4 Mini | High | 60.87 | #368 |
| 7 | GLM-4.5 | 58.59 | #542 | |
| 8 | Gemini 2.5 Pro | 55.69 | #276 | |
| 9 | Claude Opus 4.1 | 55.65 | #278 | |
| 10 | Gemini 2.5 Flash | 55.43 | #463 | |
| 11 | GPT-4o Mini | 50.88 | #1667 | |
| 12 | Qwen 2.5 72B Instruct | 50.78 | #1117 | |
| 13 | GPT-4o | 50.32 | #751 | |
| 14 | Claude Sonnet 4 | 49.84 | #410 | |
| 15 | Llama 3.3 70B Instruct | 49.44 | #1385 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=implicitmembench · How It Works · Data refreshed daily, snapshot 2026-11-21.