Home › Solutions › AI Customer Support: Cost per Ticket
Use caseAI Customer Support: Cost per Ticket
Support bots resend the same instructions and knowledge on every message. That makes them one of the best workloads for caching and routing.
Defaults use GPT-6.1 Sol's list prices ($2 input, $0.10 cached, $10 output per 1M tokens, checked October 30, 2026). Enter your own model's prices.
Where the cost goes
- A long system prompt and policies, sent on every message.
- Retrieved help articles, often more than needed.
- Conversation history that grows with each turn.
- Answers longer than customers read.
How to cut it
- Cache the system prompt and policies.
- Retrieve fewer, better chunks.
- Summarize long histories.
- Route simple intents (order status, password resets) to a cheaper model.
- Semantic caching for common questions, if the answers are the same for everyone.
Frequently asked questions
How much does AI customer support cost per ticket?
Multiply model calls per ticket by tokens per call and your model's prices. With long prompts and many turns it adds up quickly; caching usually cuts it the most.
How do I lower AI support costs?
Cache the system prompt, trim retrieved context and history, and route simple intents to a cheaper model.
Want us to find your 20%?
Book a free 15-minute API cost teardown. Bring last month's provider usage and we'll show you your biggest waste and the first fix we'd ship.
Get a Free 15-Minute API Cost Teardown