OpenAI announces significant updates for developers with new Realtime API models, a Codex Chrome extension, and advanced tools for automation and AI.
OpenAI announces significant updates for developers with new Realtime API models, Codex extension for Chrome, and advanced tools for automation and artificial intelligence.
OpenAI has released a series of updates and new tools dedicated to developers, expanding the possibilities for integration and automation based on artificial intelligence. Among the most notable innovations are real-time voice models, the Codex extension for Chrome, and the enhancement of GPT-5.5.
Real-time voice models for more natural interactionsSposta nel cestino
The new version of the Realtime API introduces three models focused on voice interactions:
- gpt-realtime-2: improves the ability to follow instructions, adhere to prompts, and supports multiple languages, for smarter and more natural voice agents.
- gpt-realtime-translate: enables simultaneous translation of conversations, supporting 70 input languages and 13 output languages.
- gpt-realtime-whisper: transcribes words as they are spoken, without waiting for the end of the sentence.
These models allow the development of voice applications that respond and act in real time, enhancing user experience in areas such as digital assistants and live translations.
Codex integrated directly into the Chrome browser
OpenAI has launched a Chrome extension that allows Codex to operate directly within websites where the user is authenticated. This integration facilitates the organization of task-specific tabs and allows results to be returned for review, expanding automation and support possibilities in daily activities performed via browser.
Other updates and tools for developers
- GPT-5.5: OpenAI’s most advanced model, with improved reasoning capabilities, more effective use of tools, and greater reliability in extended runs.
- Agents SDK: allows agents to inspect files, execute commands, modify code, and manage complex tasks in controlled environments, supporting custom sandboxes from providers such as Modal, E2B, Daytona, and Cloudflare.
- Codex: now supports over 90 new plugins and interaction features with the browser and computer environment, accelerating workflows.
- GPT-Image-2: a tool for generating professional-quality images, integrable into multimodal applications.
Resources for developers and use cases
OpenAI has also published practical guides and use cases to facilitate the adoption of new technologies, such as migrating legacy code with sandbox agents, integrating agents into official documentation, and tips to improve prompting with GPT-5.5 and Codex.
Highlighted projects include a demo of a landing page for e-commerce and a native macOS game developed with Codex and GPT-5.5, concrete examples of applying the new features.
Events and insights
OpenAI has announced the official date for DevDay 2026, an event dedicated to developers, and organizes online sessions to answer questions about new products. Additionally, it publishes technical insights on performance improvements and the architecture of real-time voice infrastructures.
Pubblicato in Artificial Intelligence
Be the first to comment