OpenAI’s New Flex Processing Offers Cheaper AI for Non-Urgent Tasks
In an effort to stay ahead in the growing competition with AI rivals like Google, OpenAI has rolled out a new offering called Flex processing — an API tier designed to make running AI models more affordable for developers who don’t need lightning-fast responses.
Announced on April 17, 2025, Flex processing is available in beta for OpenAI’s latest o3 and o4-mini reasoning models. It’s intended for non-critical and background tasks such as model testing, asynchronous operations, and data enrichment — situations where performance delays are acceptable.
Cutting Costs in Half
With Flex processing, OpenAI slashes pricing by 50%:
- For the o3 model, pricing drops to $5 per million input tokens and $20 per million output tokens (down from $10 and $40 respectively).
- For o4-mini, costs are reduced to $0.55 per million input tokens and $2.20 per million output tokens, compared to the standard $1.10 and $4.40.
However, this comes with a trade-off: slower processing speeds and potential temporary unavailability of computing resources.
Stricter Access Requirements for Developers
OpenAI also confirmed that developers falling under tiers 1 to 3 (based on spend levels) must now complete an ID verification process to gain access to the o3 model and certain features. These include streaming API capabilities and reasoning summaries, which are now gated behind verification protocols aimed at preventing misuse and abuse of the system.
This policy aligns with OpenAI’s broader goals of maintaining responsible AI usage and curbing activities from bad actors.
Competitive Pressure in the AI Space
The launch comes just as competitors make similar pricing moves. Google, for instance, recently introduced Gemini 2.5 Flash, a fast, cost-efficient reasoning model that rivals DeepSeek’s R1 in performance but at a lower input cost.
As the AI industry faces rising costs tied to running large models, pricing flexibility has become a key strategy for providers seeking to accommodate developers across different use cases.
What This Means for Developers
For developers managing high-volume or non-urgent workloads, Flex processing offers a budget-friendly option without compromising access to OpenAI’s most capable models. From app experimentation to dataset processing, it gives teams the opportunity to scale AI usage cost-effectively — albeit with some performance trade-offs.
As OpenAI continues refining its offerings, Flex could become a cornerstone for balancing efficiency, affordability, and ethical access in enterprise AI.
