Low latency matters for chat, voice and autocomplete. Reasoning models are usually slower because they think before answering. Smaller 'flash' or 'mini' models trade some quality for speed.
Frequently asked questions
What does Latency mean?
How long a model takes to respond; often measured as time to first token. Low latency matters for chat, voice and autocomplete. Reasoning models are usually slower because they think before answering. Smaller 'flash' or 'mini' models trade some quality for speed.