OpenAI Empowers Developers With GPT And Codex Updates

OpenAI is rolling out significant enhancements for its custom GPTs and the Codex code generation engine, signaling a continued effort to provide developers and power users with more sophisticated and flexible tools.

Enhanced Control for Custom GPT Builders

A highly anticipated feature for custom GPTs is on the horizon: the ability for creators to specify a recommended model for their bots. This will be facilitated through a new dropdown menu within the GPT configuration settings. Builders can select from available models, with the choice influenced by the specific features and capabilities integrated into their custom GPT. It's important to note that not all models will be universally available for every custom GPT. For instance, models highly specialized in reasoning tasks might only be selectable if the custom GPT's functionalities are compatible with such advanced capabilities.

Once implemented, end-users interacting with a custom GPT will see the builder-recommended model clearly labeled. This transparency is expected to improve user understanding and could lead to more consistent and predictable results from the custom GPTs. While the model selector feature is not yet active, its development is evidently progressing.

This update is especially beneficial for creators and businesses that utilize custom GPTs outside of OpenAI's more advanced "Projects" workflow. It grants them finer-grained control over the underlying GPT architecture, which can directly impact the quality of outputs, operational speed, and potentially the cost associated with running their bots. Given the persistent and active user base of custom GPTs, this added flexibility is poised to maintain and even enhance their value within OpenAI's expanding ecosystem of AI tools.

Information about this upcoming change was highlighted in a recent update:

New ChatGPT web app changes

• As part of the changes in the works, adding model selector for custom GPTs, GPT creators will be able to configure a recommended model (or "No Recommended Model - Users will use any model they prefer")

"Recommend a model to the user, which should… pic.twitter.com/JvheZBKziT

— Tibor Blaho (@btibor91) June 11, 2025

Codex Gets a Productivity Boost

Simultaneously, OpenAI's Codex code generation engine is set to receive a significant workflow enhancement. Users will soon have the ability to specify the number of code variations Codex should generate for any given programming request. The options will allow users to request anywhere from one to as many as eight different code solutions.

This means Codex can automatically produce multiple code drafts, a feature that will greatly streamline the development process for those looking to refine, compare, or explore alternative solutions. It effectively removes the need for developers to manually re-run tasks multiple times to obtain different coding options, saving time and effort. These improvements were identified through analysis of recent platform updates, though a specific release date has not yet been announced.

OpenAI's Strategy Driving These Enhancements

These developments for ChatGPT and its associated tools underscore OpenAI's current strategic focus: empowering developers with greater customization options while simultaneously refining its suite of products. The introduction of model recommendations for custom GPTs and automated multiple code generation attempts for Codex points towards a clear strategy of equipping power users and developers with tools that are not only more flexible but also more efficient.

Upon launch, these new features will likely be integrated into the configuration panels of custom GPTs and within Codex-powered interfaces. They are anticipated to support a broad spectrum of applications, ranging from personal automation projects to complex enterprise-level code development workflows, further solidifying OpenAI's position in the AI-assisted development landscape.