SynCred-Bench: leaderboard
Metric: True positive rate (%) on the 600 AI-generated credible-form images at an operating point whose false positive rate on the 450 real FP450 images is at most 5% (threshold swept over the stated confidence when the default FPR exceeds 5%); MLLM-as-judge detection prompt, image only. Source: arxiv.org. Saturation forecast: Around December 2026. 15 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Opus 4.6 | 69.5 |
| 2 | Claude Sonnet 4.6 | 55.3 |
| 3 | Gemini 3.1 Pro (Preview) | 12.7 |
| 4 | GPT-5.4 | 11.5 |
| 5 | Qwen 3.6 Plus | 3.7 |
| 6 | Qwen 3 VL 32B Instruct | 2.2 |
| 7 | Gemini 3.1 Flash Lite | 1.5 |
| 8 | GLM-5V Turbo | 0.5 |
| 9 | Grok 4.3 | 0.3 |
| 10 | GPT-4o | 0 |
| 11 | Qwen 3.5 9B | 0 |
| 12 | Qwen 3.5 35B A3B | 0 |
| 13 | Llama 3.2 11B Instruct | 0 |
| 14 | Qwen 3 VL 8B Instruct | 0 |
| 15 | Pixtral Large | 0 |
Interactive version: theaggregate.ai/benchmark?slug=syncred-bench · How It Works · Data refreshed daily, snapshot 2026-09-29.