ResearchGym - Cross-Modal Retrieval: leaderboard
ResearchGym task on query-shift adaptation for image-text retrieval on corrupted COCO and Flickr data. A run's score is divided by the source paper's reported result (100 = matches it) and the best of three runs counts; higher is better.
Metric: Best-of-3 score normalised to the task's reference (SOTA) result, SOTA = 100. Source: anikethh.github.io. Saturation forecast: Rough model projection: around 2027. 2 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.2 Codex (xHigh) | 96.6 | #185 (GPT-5.2 Codex) |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=researchgym-cross-modal-retrieval · How It Works · Data refreshed daily, snapshot 2026-11-12.