C3-Bench: leaderboard
While Change Captioning systems have garnered substantial attention to respond to our evolving world, their true performance on diverse real-world change contexts remains largely unexplored due to the lack of comprehensive evaluation frameworks. To fill this gap, we propose C3-Bench, a comprehensive benchmark for evaluating Context-aware Change Captioning. C3-Bench features: (1) 4,996 human-labele
Metric: Aggregation (self-reported). Source: benchmarklist.com. Saturation forecast: Around February 2027. 33 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.2 | 5.72 | #110 |
| 2 | Gemini 3 Pro | 5.5 | #80 |
| 3 | GPT-5.1 | 5.49 | #135 |
| 4 | Gemini 3 Flash | 5.39 | #99 |
| 5 | Gemini 2.5 Pro | 5.23 | #152 |
| 6 | Qwen 3 VL 8B Instruct | 5.1 | #420 |
| 7 | Gemini 2.5 Flash | 5.04 | #245 |
| 8 | GPT-4o | 4.95 | #343 |
| 9 | InternVL3-78B | 4.94 | #352 |
| 10 | GPT-5 Mini | 4.87 | #186 |
| 11 | InternVL3-38B | 4.72 | #405 |
| 12 | Llama 4 Scout | 4.71 | #656 |
| 13 | GPT-4o Mini | 4.65 | #599 |
| 14 | Qwen 2.5 VL 72B Instruct | 4.61 | #377 |
| 15 | Qwen 2.5 VL 7B Instruct | 4.41 | #654 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=c3-bench · How It Works · Data refreshed daily, snapshot 2026-10-20.