OpenAI released updated GPT-6 Sol and Luna models that offer a 50 percent price reduction compared to the previous generation, attributing the savings to inference efficiency gains. The company claims these updates halve the error rate on internal factuality evaluations while maintaining performance comparable to its top-tier Astra model. This structural shift in pricing and reliability signals a rapid compression of per-unit inference costs for enterprise-grade tasks. For financial institutions, this development intensifies margin pressure on AI software vendors who must pass through lower model costs or face customer churn.
159 other tracked developments touching the same claims.