OpenAI Introduces Flex Processing For Cost-Effective, Slower AI Operations

OpenAI Introduces Flex Processing API

OpenAI is stepping up its competition against other AI providers such as Google with the introduction of its Flex Processing API. This new option aims to offer more affordable AI model usage, albeit with some trade-offs in terms of response speed and occasional resource limits.

What is Flex Processing?

Flex Processing is currently in beta and is designed to work with OpenAI’s latest reasoning models, namely the o3 and o4-mini. This service is especially targeted at lower-priority tasks which do not require immediate responses. Examples include:

Model evaluations: Testing or assessing the performance of models.
Data enrichment: Enhancing data sets with additional information.
Asynchronous workloads: Tasks that do not require immediate interaction or output.

Cost Advantages of Flex Processing

One of the most attractive features of Flex Processing is its significant reduction in API costs. Using this new model, users will pay:

For o3:
- $5 per million input tokens (approximately 750,000 words)
- $20 per million output tokens
Traditionally, users would have to pay $10 for input tokens and $40 for output tokens, effectively halving the costs with Flex Processing.
For o4-mini:
- $0.55 per million input tokens
- $2.20 per million output tokens
This is reduced from $1.10 for input and $4.40 for output tokens in the standard model.

Market Context and Competition

The announcement of Flex Processing comes at a time when the cost of advanced AI continues to rise. Other competitors in the AI space, particularly Google, are launching more affordable and efficient models. Recently, Google introduced the Gemini 2.5 Flash, which competes directly with OpenAI’s offerings by providing comparable performance at a lower cost per token.

Accessing Flex Processing

In a communication to its users, OpenAI outlined some additional requirements for accessing the new o3 model with Flex Processing. Users in the top three tiers of OpenAI’s usage hierarchy will have to undergo a new ID verification process to continue using these models. This tier structure is based on the amount a user spends on OpenAI services.

Purpose of ID Verification

OpenAI has indicated that the intention behind implementing ID verification is to prevent misuse and ensure compliance with its usage policies. This measure aims to keep its platforms secure and mitigate the risk of bad actors exploiting the technology.