M3Exam (Multimodal Memory) - Temporal Reasoning: leaderboard
Metric: LLM-judge score (0-100) on the temporal reasoning (TR) questions of 5,150 questions over 239 long multi-session user-assistant conversations with images and PDFs (15 personas); each model answers with the annotated gold supporting sessions and their images or rendered pages placed in context, no memory module; scored by a Qwen2.5-VL-32B-Instruct judge on a five-level rubric (0, 0.25, 0.5, 0.75, 1), printed 0-1 and x100 here; higher is better. Source: arxiv.org. Saturation forecast: Around 2032. 9 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Seed 2.0 Pro | 64.37 |
| 2 | Claude Opus 4.6 | 58.13 |
| 3 | Gemini 3.1 Pro (Preview) | 54.04 |
| 4 | Grok 4 | 51.5 |
| 5 | GLM-5.1 | 51.15 |
| 6 | Qwen 2.5 VL 7B Instruct | 50.63 |
| 7 | Kimi K2.5 | 50.5 |
| 8 | GPT-5.4 | 49.5 |
| 9 | Qwen 3.6 Plus | 47.99 |
Interactive version: theaggregate.ai/benchmark?slug=m3exam-multimodal-memory-temporal-reasoning · How It Works · Data refreshed daily, snapshot 2026-09-29.