An OpenAI rival says its new AI model is not only cheaper to run than GPT-4, but it's also more useful
Cohere performed these tests itself, and found that its fine-tuned Command R model scored 80.2% on accuracy when summarizing ...