Overview

• Flex processing offers a 50% reduction in token costs for o3 and o4-mini models, targeting non-critical tasks like data enrichment and model evaluations.
• The beta service trades faster response times for cost savings, with potential timeouts and 429 errors requiring developers to implement retry strategies.
• OpenAI now requires ID verification for developers in lower spend tiers (1–3) to access o3 models and advanced features like reasoning summaries and streaming.
• Flex processing is designed for non-production workloads, providing a budget-friendly option for developers working at scale but unsuitable for real-time applications.
• This launch reflects OpenAI's strategy to compete with rivals like Google by diversifying offerings and tightening controls on access to advanced AI capabilities.