Artificial intelligence has traditionally advanced through automatic accuracy tests in tasks meant to approximate human knowledge. Carefully crafted benchmark tests such as The General Language ...

OpenAI’s o3 system scored 85% on the ARC-AGI benchmark, well above the previous AI best score of 55% and on par with the average human score. Reading time 4 minutes A new artificial intelligence (AI) ...

Pillay is an editorial fellow at TIME. Pillay is an editorial fellow at TIME. Despite their expertise, AI developers don't always know what their most advanced systems are capable of—at least, not at ...

OpenAI’s new GDPval benchmark tested GPT-5 on real-world jobs across nine industries, revealing that the AI matched or outperformed experts 40% of the time. While not a full replacement, OpenAI ...

Michael Timothy Bennett receives funding from the Australian government. Elija Perrier receives funding from the Australian government. A new artificial intelligence (AI) model has just achieved human ...

Samsung Research has launched a new AI benchmark called TRUEBench to address gaps in existing tools. The benchmark provides a more realistic evaluation of AI productivity on real-world enterprise ...

Every few weeks, you will see a tweet by some AI CEO about how their latest model has topped a benchmark. Then the headlines ...