SLVMBench - Parameter Recall: leaderboard
SLVMBench (2026) embeds a tutorial video in a stream of unrelated videos and asks video LLMs 2,261 five-option questions that apply the demonstrated skill to a target video at a precise cutoff. This board covers the parameter recall questions. Scores are accuracy in percent in the long-memory setting, with 10 minutes to 2 hours of distractor videos between the tutorial and the target; higher is better.
Metric: Accuracy (%) on the parameter recall questions of SLVMBench in the long-memory setting (tutorial, then 10 minutes to 2 hours of distractor videos, then the target video). Source: arxiv.org. Saturation forecast: Around August 2028. 9 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 57.4 | #94 |
| 2 | Qwen2.5-Omni-7B | 44.75 | #1789 |
| 3 | GPT-4o (2024-05-13) | 43.64 | #735 |
| 4 | GPT-5.2 | 41.81 | #176 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=slvmbench-parameter-recall · How It Works · Data refreshed daily, snapshot 2026-11-13.