[tag]
5 stories
A contamination-controlled benchmark of 256 tasks shows that swapping the agent framework around the same model yields statistically indistinguishable results, while cost per solved task varies significantly.
A new arXiv paper introduces a standardized framework for assessing the predictive capabilities of world models in autonomous driving scenarios.
A new open-source benchmark runs coding agents in hardened, enterprise-style sandboxes to quantify the capability cost of security controls.
A new open-source tool turns your local model collection into a live terminal leaderboard measuring speed, memory, energy, and quality in one command.
A new public leaderboard from FAR.AI aims to make frontier-model security wins and gaps more transparent.
