Tokens & Tokens per Second (t/s or tok/s)
AI models do not read or write whole words the way humans do; they break text down into chunks called tokens. In English, one token equals roughly ¾ of a word (so 100 tokens is around 75 words). Common short words are single tokens, while long technical terms, punctuation, and code indentation get split into multiple pieces.
Tokens per second (t/s or tok/s) is the primary speedometer for local AI hardware, measuring how fast a rig spits out a response:
- Under 5 t/s: Painfully slow. You watch words trickle onto the screen one by one, making long coding tasks or back-and-forth edits frustrating.
- 10–15 t/s: Comfortable reading speed. Text appears slightly faster than the average person reads, which feels totally usable for everyday chat and drafting.
- 30+ t/s: Instantaneous. Paragraphs snap onto the screen almost immediately, matching the snappy feel of paid cloud services.
Looking for verified hardware rigs?
Check our real-time £/GB VRAM index to find systems optimized for local model inference.