Jev model provides multilingual evaluation at 1/31 to 1/61 the cost of GPT-6.1

Jev is a decision model for multilingual text evaluation covering 47 language varieties. While it is significantly cheaper than GPT-6.1, it is more prone to accepting incorrect answers in low-resource languages and struggles with knowledge-intensive verification.