MemLens (128K) - Multi-Session Reasoning: leaderboard

Accuracy in percent on the MemLens 143 questions that combine evidence from several sessions, 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 multi-session reasoning (MSR, 143 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 2030. 25 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Gemini 3.1 Pro (Preview)29.37#99
2Kimi K2.528.67#266
3GLM-4.6V23.08#710
4Qwen 3.5 122B A10B22.38#346
5Phi-4 Multimodal Instruct22.38#4154
6GPT-5.421.68#135
7Qwen 3 VL 8B Instruct20.28#1029
8Cosmos-Reason2-8B20.28#755
9Qwen 3 VL 2B Instruct18.88#2367
10Gemma 3 27B (IT)18.18#1324
11Gemma 3 12B (IT)18.18#1887
12Qwen 3 VL 235B A22B Instruct18.18#532
13Qwen 3 VL 4B Instruct16.08#1332
14Qwen 3 VL 30B A3B Instruct16.08#895
15Qwen 3 VL 235B A22B (Thinking)15.38#459

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-multi-session-reasoning · How It Works · Data refreshed daily, snapshot 2026-11-21.