MemLens (32K) - Information Extraction: leaderboard
Metric: Accuracy (%) on the information extraction (IE, 246 questions) at a 32K-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 February 2028. 27 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Qwen 3.5 122B A10B | 74.39 |
| 2 | Qwen 3.5 27B | 70.33 |
| 3 | GPT-5.4 | 69.51 |
| 4 | Qwen 3.5 9B | 67.48 |
| 5 | Qwen 3 VL 30B A3B Instruct | 65.45 |
| 6 | Qwen 3.5 2B | 64.23 |
| 7 | Nemotron Nano 12B v2 VL | 63.82 |
| 8 | GLM-4.5V | 61.79 |
| 9 | Qwen 3 VL 235B A22B Instruct | 60.98 |
| 10 | GLM-4.6V | 59.35 |
| 11 | Gemini 3.1 Pro (Preview) | 57.32 |
| 12 | Qwen 3.5 4B | 56.91 |
| 13 | Qwen 3 VL 8B Instruct | 52.85 |
| 14 | Qwen 3 VL 4B Instruct | 52.03 |
| 15 | Qwen 3 VL 235B A22B (Thinking) | 51.63 |
Interactive version: theaggregate.ai/benchmark?slug=memlens-32k-information-extraction · How It Works · Data refreshed daily, snapshot 2026-10-07.