Episodic Memory: leaderboard
Tests how well LLMs encode, store, and recall episodic events across long narratives (200 chapters, ~100K tokens, 686 Q&A pairs). Measures simple recall and chronological awareness.
Metric: Simple Recall (%). Source: github.com. Saturation forecast: Estimated already saturated. 21 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 96.8 | #152 |
| 2 | Gemini 2.5 Flash | 96 | #245 |
| 3 | GPT-5 | 94.2 | #98 |
| 4 | GPT-5 Mini | 83 | #186 |
| 5 | Claude Sonnet 4 | 79 | #199 |
| 6 | Grok 4 Fast (Reasoning) | 72.6 | #254 (Grok 4 Fast) |
| 7 | Gemini 2.0 Flash (Thinking) | 70.8 | #342 (Gemini 2.0 Flash) |
| 8 | GPT-4o | 67 | #343 |
| 9 | Grok 4 Fast (Non-reasoning) | 60.2 | #254 (Grok 4 Fast) |
| 10 | DeepSeek V3 | 60 | #321 |
| 11 | Gemini 2.0 Flash | 59.6 | #342 |
| 12 | DeepSeek R1 | 57.2 | #251 |
| 13 | Llama 3.1 405B | 50.4 | #466 |
| 14 | GPT-4o Mini | 49.2 | #599 |
| 15 | Claude 3.5 Sonnet | 47 | #350 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=episodic-memory · How It Works · Data refreshed daily, snapshot 2026-10-20.