OpenAI has introduced a new API option called Flex processing, aimed at delivering lower-cost access to its advanced AI models by offering slower response times and occasional resource limitations. The new feature specifically targets tasks that are less time-sensitive or considered non-critical, such as asynchronous workloads, data enrichment, and model evaluations.
Currently available in beta, Flex processing applies to OpenAI’s recently launched o3 and o4-mini reasoning models, providing a significant reduction—exactly 50% off standard API pricing. For example, utilizing Flex processing on the o3 model brings costs down to $5 per million input tokens (approximately 750,000 words) and $20 per million output tokens, compared to regular rates of $10 per million input tokens and $40 per million output tokens. Similarly, the o4-mini model, under the Flex option, costs $0.55 per million input tokens and $2.20 per million output tokens, down from the standard rates of $1.10 and $4.40, respectively.
OpenAI’s introduction of Flex processing comes as the cost of utilizing advanced AI technology continues to rise and as competitors, such as Google, introduce alternative models emphasizing lower costs and improved efficiency. Google recently rolled out its Gemini 2.5 Flash reasoning model, offering competitive performance at reduced input token pricing compared to other market leaders.
Alongside the launch of Flex processing, OpenAI announced through customer notifications that developers categorized into tiers 1 through 3 of its usage-level hierarchy will need to complete the company’s new identification verification process to gain access to the powerful o3 model. Additionally, ID verification is now also required to access certain o3 features, including reasoning summaries and API streaming capabilities. OpenAI previously stated that instituting an identity verification protocol is essential for preventing misuse of its platform by malicious actors.
