With new speech models in the OpenAI API, real-time conversation, live translation, and transcription are moving closer to everyday workflows — audio no longer has to wait until after the fact to be processed. Interesting for any team looking to reduce the manual overhead around support calls, training sessions, or international meetings. Here’s what the models can do and where

GPT-5.5 brings finer control over how deeply ChatGPT thinks — swapping a one-size-fits-all mode for a deliberate choice between speed and depth. Particularly useful for anyone who regularly switches between quick replies and complex analysis. Here’s what’s changed, how to use it, and where it actually makes a difference.

ChatGPT Images 2.0 takes image generation to the next level — visuals are created and refined directly in the conversation, no separate tools required. The workflow is simple: describe what you need, get an image, adjust as you go. For presentations, internal communications, or quick mockups, that kind of frictionless iteration can make a real difference. Here’s what’s new, how