OpenAI launches Flex processing for cheaper, slower AI tasks
In an attempt to compete more aggressively with rival AI companies such as Google, OpenAi is launched Flex processingAn API option that offers lower AI model use prices in exchange for slower response times and “incidental resources not available”.
Flex processing, which is available in BETA for the recently released O4-mini reasoning models of OpenAi, is aimed at tasks with a lower priority and “non-production” such as model evaluations, data enrichment and asynchronous workload, says OpenAI.
It lowers API costs by exactly half. For O3, Flex processing is $ 5/m input tokens (~ 750,000 words) and $ 20/m export tokens versus the standard $ 10/m input sticks and $ 40/m export tokens. For O4-Mini, Flex brings the price to $ 0.55/m input sticks and $ 2.20/m output tokens of $ 1.10/m input sticks and $ 4.40/m output tokens.
The launch of flex processing comes if the price of Frontier AI continues to rise, and while rivals release cheaper, more efficient budget-oriented models. Google rolled out on Thursday Gemini 2.5 FlashA reasoning model that corresponds to the R1 of the Deepseek in terms of performance at a lower input token costs.
In one E -Mail to customers OpenAI announced the launch of flex prices and also indicated that developers in levels 1-3 of his user archion have to complete the newly introduced ID verification process to gain access to O3. (Layers are determined by the amount of money spent on OpenAI services.) O3s – and other models – revised summaries and streaming API support also become rear verification.
OpenAi said earlier that ID verification is intended to prevent bad actors from violating his user policy.
