OpenAI Launches Flex Processing: Cheaper AI Model Access for Slower, Non-Critical Tasks

In an aggressive push to make its AI models more accessible — and to fend off intensifying competition from Google and other rivals — OpenAI has unveiled a new cost-saving option for developers: Flex Processing.

Announced today, Flex Processing allows customers to access OpenAI’s powerful AI models at half the usual price by accepting longer response times and the possibility of occasional resource unavailability. The option is currently in beta and available for the company’s latest models, o3 and o4-mini.

OpenAI says Flex Processing is ideal for lower-priority, non-production workloads such as:

Model evaluations
Data enrichment
Asynchronous background tasks

For developers, the savings are significant. Flex cuts API costs by 50% compared to standard pricing:

For o3, Flex drops input token pricing to $5 per million (about 750,000 words) and output token pricing to $20 per million — down from the standard $10 and $40 rates.
For o4-mini, the Flex rate is $0.55 per million input tokens and $2.20 per million output tokens, compared to the usual $1.10 and $4.40.

The launch couldn’t come at a more strategic moment. As the AI arms race heats up, companies are feeling the squeeze of rapidly rising compute and inference costs. OpenAI’s move follows Google’s release of Gemini 2.5 Flash, a new reasoning model designed for high efficiency at lower prices, directly rivaling offerings like DeepSeek’s R1.

In addition to Flex, OpenAI used the announcement to remind developers that higher-tier customers (tiers 1-3) will need to complete a new ID verification process to maintain access to models like o3. This step, the company says, is part of its ongoing efforts to prevent misuse and ensure responsible deployment of AI technology.

The verification requirement will also apply to features such as reasoning summaries and streaming API support.

As competition stiffens and customers become more cost-conscious, Flex Processing positions OpenAI to retain budget-sensitive developers who still want access to state-of-the-art models, just without the production-grade service level.

Get the Latest AI News on AI Content Minds Blog