The company claims it secured a top score of 89.4 in the Arena-Hard benchmark, which compares AI models based on how they respond to human prompts.
The company claims it secured a top score of 89.4 in the Arena-Hard benchmark, which compares AI models based on how they respond to human prompts.