Back to glossary

Term

Tier Pricing

Tiered model line-up from a provider — small fast variants (Mini/Flash/Haiku) at a fraction of the price of the big frontier models. Also: volume tiers with quantity discounts.

Tier pricing — in more detail

Providers stagger their models into tiers so users can pick the right price-performance level for each task. The rule of thumb: frontier models (Opus, GPT-5.5, Gemini 3 Pro) for complex reasoning and edge cases; mid-tier (Sonnet, GPT-5.5-mini, Gemini Flash) for most production workloads; low-tier (Haiku, GPT-Nano, Flash-Lite) for high-volume, simple tasks like classification or routing. The price gap between top and bottom tier is typically a factor of 30–60.

Example / practical context

A chatbot that first classifies incoming tickets (short, simple task) and then produces a detailed reply depending on the class can run the classification step on Haiku (~$0.25 / $1.25 per MTok) and only the final answer on Sonnet (~$3 / $15). At 10,000 tickets per day this often saves 70–90 % of total cost versus a pure Sonnet setup — with almost no quality loss, because classification simply is not a frontier-class task.

Volume tiers (also: usage tiers) are a different axis — they stagger discounts by monthly spend or prepayment, not by model size. Tier pricing here means model selection. Combined they yield the cheapest setup: small model for simple tasks + high volume tier for quantity discount.

Entdecke mehr