EverMemBench (Full Context) - Fine-Grained Recall (Single-Hop): leaderboard
EverMemBench (Hu et al., 2026) tests long-horizon memory over simulated multi-party workplace chat: five year-long projects, about 845,000 tokens of group dialogue each, and 2,400 questions, here with the whole dialogue history in the model's context. Recalling one precise fact, such as a link, name or date, among near-duplicate distractors; open-ended answers judged right or wrong by an LLM, accuracy in percent, higher is better.
Metric: Accuracy (%; whole multi-party dialogue history in context, open-ended, judged by an LLM). Source: arxiv.org. Saturation forecast: Estimated already saturated. 3 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3 Flash | 97.65 | #159 |
| 2 | GPT-4.1 Mini | 83.57 | #775 |
| 3 | Llama 4 Scout Instruct | 77.93 | #1532 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=evermembench-full-context-fine-grained-recall-single-hop · How It Works · Data refreshed daily, snapshot 2026-11-12.