Benchmarking: We used WMT26 benchmarks, which were released after we created the model, ensuring we couldn't train North Small Translate to its measurements. (5/6) pic.twitter.com/amZsHNndd7
— Cohere (@cohere) 3 octobre 2026
Benchmarking: We used WMT26 benchmarks, which were released after we created the model, ensuring we couldn't train North Small Translate to its measurements. (5/6)