Imaging-101 - End-to-End: leaderboard
Imaging-101 (2026) tests LLM coding agents on 57 computational imaging tasks from six scientific domains, each grounded in a peer-reviewed paper. This board is the percent of tasks whose full reconstruction pipeline, written and run by the agent, meets the image-quality thresholds of the task; higher is better.
Metric: End-to-end success rate (%): share of the 57 expert-verified computational imaging tasks from six scientific domains whose full reconstruction pipeline, written and executed by the agent, meets the task's NCC and NRMSE image-quality thresholds; higher is better. Source: arxiv.org. Saturation forecast: Around November 2027. 7 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Opus 4.6 | 29.8 | #102 |
| 2 | GPT-5.4 | 19.3 | #123 |
| 3 | Gemini 3.1 Pro (Preview) | 14 | #94 |
| 4 | GLM-5 | 12.3 | #260 |
| 5 | DeepSeek V3 | 10.5 | #716 |
| 6 | Kimi K2.5 | 7 | #256 |
| 7 | Qwen 3.6 Plus | 7 | #222 |
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-end-to-end · How It Works · Data refreshed daily, snapshot 2026-11-13.