A new research paper from Apple details a technique that speeds up large language model responses, while preserving output quality. Here are the details.

Traditionally, LLMs generate text one token at a time. This is slow because each step depends on all the previous ones to keep the output coherent and accurate.

If the model is writing a sentence like “ The cat is black ”, it predicts each token in sequence. After writing “ The cat is ”, it looks at everything so far (plus the user’s

Yesterday OpenAI released GPT-5 , the latest model update to ChatGPT. More than 24 hours later, the new model is still rolling out to paid and free customers. Meanwhile, OpenAI is reconsidering the availability of certain models and limitations for customers. The new model will also be more transparent and offer more control over which tools it uses for which requests.

Let’s start with the rollout status. I was pleasantly surprised when OpenAI said GPT-5 would be available to almost all of its users starting yesterday