Imaging-101 - Planning: leaderboard
The planning track of Imaging-101 (2026): the percent of its 57 computational imaging tasks whose written plan gets preprocessing, forward physics and the inverse solver all correct, as judged by domain experts. Higher is better.
Metric: Planning pass rate (%): share of the 57 expert-verified computational imaging tasks from six scientific domains whose written plan gets preprocessing, forward physics and inverse solver all correct (two blinded domain experts reviewing an LLM-judge difference report); higher is better. Source: arxiv.org. Saturation forecast: Around 2030. 7 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.4 | 52.6 | #123 |
| 2 | Kimi K2.5 | 52.6 | #256 |
| 3 | Claude Opus 4.6 | 50.9 | #102 |
| 4 | DeepSeek V3 | 49.1 | #716 |
| 5 | Qwen 3.6 Plus | 45.6 | #222 |
| 6 | Gemini 3.1 Pro (Preview) | 40.4 | #94 |
| 7 | GLM-5 | 38.6 | #260 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=imaging-101-planning · How It Works · Data refreshed daily, snapshot 2026-11-13.