Some features may increase costs by applying a multiplier to the base token pricing:RAG 1.0×, STM 1.05×, LTM 1.10×, Personalization 1.25×.
RAG: Retrieval-Augmented Generation, STM: Short-term Memory, LTM: Long-term Memory. Personalization uses profile/context analysis.
If you always used gpt-5-nano with the highest personalisation level (1.25×), $10 would give you about 10,456 complete message–reply pairs (about 10 million tokens total). This assumes each exchange is around 300 tokens in your message and 700 tokens in the reply. Shorter messages give you even more turns; longer ones give fewer.
Your chat will pause and we'll show a simple "Add credits" prompt—no surprises.