DeepSeek launched its V4-Pro mannequin at considerably larger worth factors than its Flash variant, with output token prices reaching about $3.96 per million tokens throughout peak utilization, in contrast with a lot decrease charges for its V4-Flash mannequin.
“Token cost has been the practical ceiling on scaling AI beyond isolated pilots,” Chandak mentioned. At cheaper price factors, he added, operating agent-based workflows at manufacturing scale turns into extra viable.
He additionally mentioned enterprises are putting better emphasis on elements past mannequin efficiency. “The base model layer is commoditizing,” Chandak mentioned, including that differentiation will more and more rely on information readiness, governance, and orchestration layers.
