Which Gemini tier gives the lowest cost per message at this volume?

Google positions Flash-Lite as its most cost-efficient model, optimized for low-latency, high-volume, cost-sensitive traffic - which is exactly what classifying six million short messages a day is. Flash is the close call worth knowing: it is also fast and also built for volume, but it sits a tier above Flash-Lite on price, so at this scale the difference is the whole budget. Pro tiers are positioned for complex reasoning and coding and cost more per message than either. The Live API is built for real-time bidirectional audio and video conversation; the word 'latency' appears in its description, but nothing here is a live conversation.