MemLens (128K) - Knowledge Update: leaderboard

Accuracy in percent on the MemLens 116 questions whose answer changed in a later session, at a 128K-token interleaved image and text conversation history, with the evidence sessions hidden among distractor sessions (MemLens, 2026), judged by Qwen3-VL-235B. Higher is better.

Metric: Accuracy (%) on the knowledge update (KU, 116 questions) at a 128K-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 2027. 22 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1GPT-5.443.1#135
2Qwen 3.5 27B38.79#375
3Qwen 3 VL 235B A22B Instruct30.17#532
4Qwen 3 VL 235B A22B (Thinking)30.17#459
5GLM-4.6V28.45#710
6Qwen 3.5 9B25.86#828
7Qwen 3 VL 30B A3B Instruct24.14#895
8Qwen 3 VL 8B (Thinking)24.14#910
9Qwen 3 VL 30B A3B (Thinking)20.69#719
10Qwen 3.5 4B19.83#1241
11Claude Sonnet 4.516.38#261
12Gemma 3 12B (IT)16.38#1887
13Qwen 3 VL 8B Instruct15.52#1029
14Phi-4 Multimodal Instruct14.66#4154
15Gemma 3 27B (IT)13.79#1324

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=memlens-128k-knowledge-update · How It Works · Data refreshed daily, snapshot 2026-11-21.