• token — The chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.
• reasoning model — A model trained to think — generating extended internal reasoning before answering — trading time and tokens for accuracy on hard problems.

Voice agents still run on GPT-4.1 because no reasoning model fits a 500ms budget, and that model is scheduled to sunset on many providers. Parallel-refinement decoding is the argued escape route without waiting on scarce custom silicon.

• articleRise Of Realtime Subagents
• articleUltra Fast Apply Edit With Mercury Coder
• articleMercury Refreshed
• articleIntroducing Mercury

• articleIntroducing Mercury 2Inception editorial sitemap
• postVoice AI Fast-Path Architecture UpdateOpenAI
• clipTool calling in a real-time voice model: two models, not oneJulia Turc

Checking sign-in…

Loading comments…