OpenAI introduces GPT-5 Pro, new models in its API, and updates Codex at DevDay 2025

OpenAI held its DevDay on Monday, October 6, an event specifically aimed at developers where the technology company introduced numerous innovations through live conferences and demonstrations: from the announcement of its most cutting-edge model, GPT-5 Pro, to the general availability of Codex and the integration of new models into the API.

During the event, Sam Altman, CEO of OpenAI, stated that 4 million developers have already used OpenAI’s technology to develop their own solutions. The company’s goal with this new round of updates is to continue attracting these professionals and ensure its products help them further transform the future of AI.

Additionally, another highly revealing figure that Altman shared with the audience was the following: ChatGPT now has more than 800 million weekly active users.

GPT-5 Pro in the API

The GPT-5 Pro model was one of the major announcements at DevDay. This is the most advanced version of OpenAI’s GPT-5 language model, designed for complex and professional tasks where accuracy is critical. Because this AI uses more computing resources in order to “think” more deeply and deliver more consistent responses, this slows down its response times in comparison to other versions.

Unlike previous model releases, in this instance, GPT-5 Pro has only been activated in the API, and is reserved for a specific type of interaction within its infrastructure: the “Responses API.”

“GPT-5 pro is available in the Responses API only to enable support for multi-turn model interactions before responding to API requests, and other advanced API features in the future,” explains OpenAI.

Altman explained that GPT-5 Pro could be an excellent option, particularly for those developers who create applications that require a high degree of precision and depth of reasoning. For example, in sectors such as finance, legal, or healthcare.

More updates in the API

The technology company took this opportunity to integrate new models into the API, enhancing its appeal. These are the main updates:

• GPT-realtime mini: This is the reduced version of GPT-realtime. It is a voice model that is 70% more cost-effective and is capable of responding to both audio and text input in real time via WebRTC, WebSocket, or SIP connections.
• GPT-image-1-mini: This is the smallest version of GPT-image-1, designed for the API. It is a native multimodal language model that accepts both text and image input and produces image outputs. OpenAI states that it is 80% less expensive than the larger model.
• Sora 2: This is the most powerful version of OpenAI’s AI for generating video and audio. Developers will be able to access and use this model to power their applications.

Codex reinvents itself

OpenAI introduced Codex, its cloud-based software engineering agent capable of working on many tasks in parallel, in May 2025, and has now announced its general availability as well as new features. Whether you program in your editor, terminal, or in the cloud, Codex is becoming a more reliable and competent collaborator.

• Integration with Slack: You can now tag @Codex directly in a Slack channel or thread. The advantage is that Codex not only replies to you, but also automatically captures the context of the conversation, selects the appropriate work environment, and returns a link to the completed task in the Codex cloud. From there, you decide: merge the changes, continue iterating, or download the code to work locally.
• Codex SDK: This allows you to incorporate the same agent that powers the Codex CLI with just a few lines of code. This ensures fast, accurate results without the need for further adjustments. It also provides context management so that you may resume sessions and it is already available for TypeScript (and soon for more languages). In addition, they have launched a GitHub Action to facilitate its use in CI/CD pipelines.
• New administration features: These give administrators of ChatGPT workspaces much greater visibility and control. Now, they can edit or delete environments in the Codex cloud, define safer default settings for local use, and also have access to entirely new analytics dashboards. These dashboards enable the monitoring of usage in the CLI, IDE, and web, and even allow you to track the quality of code reviews suggested by Codex. This is a significant step toward ensuring control and scalability.

Photo: DevDay