The system will now account for the actual computational complexity of queries: the volume of models used, context length, and the complexity of functions executed. Limits will reset every five hours. This is a pragmatic macroeconomic maneuver. The cost of maintaining GPU clusters is forcing Big Tech to tighten the screws: the era of "unlimited" AI assistants is over. Corporations will force users (both B2C and B2B) to pay for every processor cycle "burned." This trend towards dynamic pricing will soon become an industry standard, forcing developers to carefully optimize their prompts.
Source: Google Support
PricingGoogleGeminiUnit EconomicsMacroeconomics