* Message estimates are based on average model costs and typical usage (~1000 tokens per conversation). Actual usage may vary depending on message length, chosen model, and conversation context. See our Terms of Service for detailed usage calculation.