ChronoVision - News Image-Text Matching: leaderboard
News image-text matching on ChronoVision (2026): in 1,500 tests the model picks which of four photos from different years matches the year of a described historical news event, and the score is the percent answered correctly. Higher is better.
Metric: Accuracy (%) on the 1,500 image-text tests: pick which of four photos from different years matches the year of a described historical news event; zero-shot; higher is better. Source: arxiv.org. Saturation forecast: Around February 2027. 10 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 2.5 Pro | 84.27 | #266 |
| 2 | GPT-5.2 | 49.73 | #176 |
| 3 | GLM-4.1V-9B (Thinking) | 38.27 | #1060 (GLM-4.1V-9B) |
| 4 | Qwen 3 VL 235B A22B Instruct | 33.8 | #522 |
| 5 | Qwen 3 VL 4B Instruct | 26.33 | #1279 |
| 6 | Qwen 3 VL 8B Instruct | 26.13 | #981 |
| 7 | Qwen 2.5 VL 7B Instruct | 12.47 | #1833 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=chronovision-news-image-text-matching · How It Works · Data refreshed daily, snapshot 2026-11-13.