IKEA-Bench - Next-Step Prediction: leaderboard
Metric: Accuracy (%) of next-step prediction (pick the diagram of the step after the one shown in video from four diagrams; 204 questions; chance 25) on IKEA-Bench, wordless IKEA manual diagrams against real assembly video from 29 products, Visual setting (diagram images only), zero-shot greedy decoding, unparsed answers counted wrong; higher is better. Source: arxiv.org. Saturation forecast: Around 2033. 19 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3 Flash | 43.1 |
| 2 | Gemini 3.1 Pro (Preview) | 41.7 |
| 3 | Qwen 3.5 27B | 41.2 |
| 4 | Qwen 3.5 9B | 38.2 |
| 5 | Qwen 2.5 VL 7B | 36.8 |
| 6 | Qwen 3.5 2B | 36.3 |
| 7 | GLM-4.1V-9B (Thinking) | 35.8 |
| 8 | Gemma 3 27B | 31.4 |
| 9 | Gemma 3 4B | 29.4 |
| 10 | Gemma 3 12B | 28.4 |
Interactive version: theaggregate.ai/benchmark?slug=ikea-bench-next-step-prediction · How It Works · Data refreshed daily, snapshot 2026-10-07.