AgentDish directory
preprint
Accepted listings with this tag.
| Listing | Category | Score | Trend | Checked | |
|---|---|---|---|---|---|
A Zenodo preprint reporting a 12,160-trial black-box evaluation of GPT-5.4 outputs under two prompt conditions, with detailed token-ceiling sweeps, control/ablation trials, and hash-chained verification. |
Research / AI Model Evaluation | 72 | ↑ +1 | 26 days ago | Details |