C3-Bench - Correctness: leaderboard
C3-Bench (2026) evaluates context-aware change captioning: describing what changed between two images according to criteria that differ across 51 real-world change contexts. This board is correctness (0 to 10), how closely the caption matches the reference change, scored by a GPT-5.2 judge over 4,996 human-labelled image pairs and averaged over three judge runs. Higher is better.
Metric: Correctness (how closely the caption matches the reference change) (0-10) of change captions on the 4,996 human-labelled C3-Bench image pairs (51 change contexts across natural scenes, remote sensing, image editing and anomalies), scored by a GPT-5.2 judge against the reference description with the context-specific change criteria; mean of three judge runs; higher is better. Source: arxiv.org. Saturation forecast: Around May 2027. 26 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.2 | 5.47 | #176 |
| 2 | Gemini 3 Pro | 5.45 | #137 |
| 3 | GPT-5.1 | 5.43 | #229 |
| 4 | Gemini 3 Flash | 5.42 | #164 |
| 5 | Gemini 2.5 Pro | 5.18 | #266 |
| 6 | Qwen 3 VL 32B Instruct | 5.18 | #585 |
| 7 | Qwen 3 VL 8B Instruct | 5.07 | #981 |
| 8 | Gemini 2.5 Flash | 4.99 | #453 |
| 9 | GPT-4o | 4.95 | #741 |
| 10 | GPT-5 Mini | 4.87 | #348 |
| 11 | InternVL3-78B | 4.86 | #738 |
| 12 | InternVL3-38B | 4.72 | #903 |
| 13 | Llama 4 Scout Instruct | 4.63 | #1543 |
| 14 | GPT-4o Mini | 4.56 | #1623 |
| 15 | Qwen 2.5 VL 72B Instruct | 4.54 | #846 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=c3-bench-correctness · How It Works · Data refreshed daily, snapshot 2026-11-13.