Hindsight LLM Memory Leaderboard: leaderboard
LLM leaderboard for Hindsight agent-memory operations, measuring retain(), reflect(), and quality performance over memory extraction and recall workloads.
Metric: Quality Accuracy (percentage). Source: github.com. Source: benchmarklist.com. Saturation forecast: Estimated already saturated. 17 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5 Mini | 89.7 | #186 |
| 2 | GPT-4.1 Nano | 87.2 | #731 |
| 3 | GPT-5.4 | 86.8 | #79 |
| 4 | GPT-4.1 Mini | 86.4 | #355 |
| 5 | GPT-5.4 Mini | 86.4 | #212 |
| 6 | Gemma 4 31B | 86 | #200 |
| 7 | Gemini 2.5 Flash | 85.5 | #245 |
| 8 | GPT-OSS-120B | 84.7 | #341 |
| 9 | Gemini 2.5 Flash Lite | 84.7 | #422 |
| 10 | GPT-5 Nano | 83.9 | #425 |
| 11 | GPT-5.4 Nano | 83.9 | #328 |
| 12 | GPT-5.2 | 83.5 | #110 |
| 13 | Gemini 3 Flash (Preview) | 83.5 | #82 |
| 14 | GPT-4o Mini | 81 | #599 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=hindsight-llm-memory-leaderboard · How It Works · Data refreshed daily, snapshot 2026-10-20.