SFI-Bench - Operation Planning (Web Search): leaderboard
Metric: Accuracy (%) on the operation planning questions, four-option multiple choice on egocentric indoor videos (ARKitScenes and ScanNet++ scans), zero-shot with the same prompt templates, models at their default configurations; the model may first retrieve external knowledge such as user manuals through its web search tool; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 8 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 3.1 Flash Lite | 77.7 |
| 2 | Gemini 3.1 Pro (Preview) | 67.9 |
| 3 | GPT-5.4 (High) | 65.5 |
| 4 | O4 Mini | 65 |
| 5 | GPT-5.4 | 63.4 |
| 6 | Gemini 2.5 Pro | 60.2 |
| 7 | GPT-5 | 60.2 |
| 8 | Gemini 2.5 Flash | 50.4 |
Interactive version: theaggregate.ai/benchmark?slug=sfi-bench-operation-planning-web-search · How It Works · Data refreshed daily, snapshot 2026-10-07.