MemLens (128K) - Temporal Reasoning: leaderboard
Accuracy in percent on the MemLens 194 questions about dates, durations and the order of events, 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 temporal reasoning (TR, 194 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 August 2027. 24 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3 VL 235B A22B Instruct | 54.64 | #532 |
| 2 | Qwen 3 VL 30B A3B Instruct | 52.58 | #895 |
| 3 | Qwen 3 VL 8B Instruct | 44.85 | #1029 |
| 4 | GLM-4.6V | 43.81 | #710 |
| 5 | Cosmos-Reason2-8B | 43.81 | #755 |
| 6 | Qwen 3 VL 235B A22B (Thinking) | 43.3 | #459 |
| 7 | Qwen 3.5 122B A10B | 41.75 | #346 |
| 8 | Gemini 3.1 Pro (Preview) | 41.24 | #99 |
| 9 | GPT-5.4 | 39.69 | #135 |
| 10 | Qwen 3 VL 8B (Thinking) | 39.18 | #910 |
| 11 | Gemma 3 27B (IT) | 37.11 | #1324 |
| 12 | Qwen 3 VL 30B A3B (Thinking) | 36.6 | #719 |
| 13 | Qwen 3 VL 4B Instruct | 35.05 | #1332 |
| 14 | Qwen 3.5 27B | 34.02 | #375 |
| 15 | Claude Sonnet 4.5 | 32.99 | #261 |
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-temporal-reasoning · How It Works · Data refreshed daily, snapshot 2026-11-21.