AI sector sees new developments in model evaluation and open-source data

Vals AI has secured $40 million in Series A funding, led by Andreessen Horowitz, to develop independent evaluation benchmarks for artificial intelligence models. The company focuses on assessing how large language models perform in economically significant tasks, such as financial analysis, legal research, coding, and web search, rather than abstract academic puzzles. Its flagship product, the Vals Index, ranks models based on real-world utility to assist enterprise buyers in procurement decisions.

In a separate development, the OpenWALDO project has been launched to create a shared, open-source AI training dataset. Led by Gregory Kurtzer, founder of CentOS and Rocky Linux, and funded by CIQ, the initiative aims to increase transparency in AI model design. OpenWALDO seeks to address concerns regarding the use of closed-source, copyrighted, or unconsented data in current open-weight models by fostering a collaborative, open-source approach to training data.