MemLens (64K) - Information Extraction: leaderboard
Metric: Accuracy (%) on the information extraction (IE, 246 questions) at a 64K-token history, interleaved multimodal multi-session conversation history with the evidence sessions hidden among haystack sessions, judged correct or incorrect by a Qwen3-VL-235B-A22B-Instruct judge; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 27 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3.5 122B A10B | 67.07 |
| 2 | GPT-5.4 | 63.01 |
| 3 | Qwen 3.5 27B | 62.2 |
| 4 | GLM-4.6V | 61.79 |
| 5 | Qwen 3.5 9B | 60.57 |
| 6 | Gemini 3.1 Pro (Preview) | 58.94 |
| 7 | Qwen 3 VL 30B A3B Instruct | 56.91 |
| 8 | Qwen 3 VL 235B A22B Instruct | 56.1 |
| 9 | Nemotron Nano 12B v2 VL | 56.1 |
| 10 | Kimi K2.5 | 52.44 |
| 11 | Qwen 3.5 4B | 49.19 |
| 12 | GLM-4.5V | 45.93 |
| 13 | Qwen 3 VL 8B Instruct | 44.72 |
| 14 | Qwen 3 VL 4B Instruct | 42.28 |
| 15 | Qwen 3 VL 8B (Thinking) | 42.28 |
Interactive version: theaggregate.ai/benchmark?slug=memlens-64k-information-extraction · How It Works · Data refreshed daily, snapshot 2026-10-07.