OpenAI updates GPT-4o in ChatGPT with better coding and fewer emojis

OpenAI just quietly pushed another update to GPT-4o inside ChatGPT.

There was no big keynote or version bump this time, just a quick post on X detailing four practical refinements.

Better prompt following, fewer emojis

According to OpenAI, the updated GPT-4o is significantly better at following detailed instructions, especially when you pack multiple requests into a single prompt.

Technical problem-solving and coding capabilities also received an upgrade, alongside what the company describes as improved intuition and creativity.

And for anyone tired of chatty bot banter, the model will now use noticeably fewer emojis.

Why it matters

We think OpenAI is addressing the exact friction points that annoy everyday power users the most.

When you use an AI for real work, few things are more frustrating than having the model ignore your third constraint or clutter technical output with smiley faces. Anyone relying on multi-step prompts or code debugging should feel an immediate improvement without changing how they work.

It also highlights how frontier models are being maintained like mature software products. You do not need a headline-grabbing new model every month; quietly eliminating instruction drift makes the assistant substantially more dependable.