Gemini 3.8 Flash's unchanged per-tokenThe chunk of text a model reads and writes in — roughly three-quarters of a word — and the unit AI usage is billed in.Full definition → price does not imply an unchanged bill for each completed task. Google DeepMind says the model performs more reasoning steps and repeatedly calls tools on difficult work, potentially consuming more tokens at higher effort settings. Its September 2 announcement presents that extra work as part of its coding and reasoning gains.
Developers can choose lower effort levels when consumption matters more, or retain 3.7 Flash for efficiency-focused workloads. The useful comparison is therefore the quality and total cost of a completed task at a chosen effort level, with per-token pricing only one part of that calculation.
Terms in this piece · Glossary
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.
