MemLens (128K): leaderboard
Metric: Overall accuracy (%) on all 789 questions (information extraction, multi-session, temporal and knowledge-update reasoning and answer refusal) 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 January 2027. 25 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 51.99 |
| 2 | Kimi K2.5 | 51.99 |
| 3 | GPT-5.4 | 49.56 |
| 4 | Qwen 3 VL 235B A22B Instruct | 47.66 |
| 5 | Qwen 3.5 122B A10B | 45.5 |
| 6 | Qwen 3 VL 30B A3B Instruct | 44.23 |
| 7 | Qwen 3.5 27B | 40.68 |
| 8 | GLM-4.6V | 39.04 |
| 9 | Qwen 3 VL 8B Instruct | 34.73 |
| 10 | Qwen 3 VL 235B A22B (Thinking) | 34.73 |
| 11 | Qwen 3.5 9B | 32.57 |
| 12 | Qwen 3 VL 30B A3B (Thinking) | 31.31 |
| 13 | Qwen 3 VL 8B (Thinking) | 30.29 |
| 14 | Qwen 3 VL 4B Instruct | 28.14 |
| 15 | Claude Sonnet 4.5 | 27.76 |
Interactive version: theaggregate.ai/benchmark?slug=memlens-128k · How It Works · Data refreshed daily, snapshot 2026-10-07.