What does a email triage agent cost to run?
An agent that reads inbound email, classifies and routes it, and drafts a reply. High volume, short inputs and outputs, one tool call to apply a label or create a ticket. On GPT-4o mini this works out to about $123 a month; here is the figure across every model and what drives it.
assumptions
A planning estimate for this shape of workload. Tune any of it in the calculator.
- 6,000 emails a day
- ~900 input tokens per email
- ~250 output tokens (label plus draft)
- 1 tool call (route or create ticket)
monthly_cost · GPT-4o mini
$123/ month
- Input tokens2.5k/req · agentic context
- $68.85
- Output tokens500/req
- $54.00
2.5k input · 500 output · 2 LLM turns / request
cost_by_model
A email triage agent across every model
| model | cost / month |
|---|---|
| Gemini 1.5 FlashGoogle (Vertex) | $61.43cheapest |
| GPT-4o miniOpenAI · shown above | $123 |
| Claude HaikuAnthropic | $727 |
| Gemini 1.5 ProGoogle (Vertex) | $1,024 |
| GPT-4oOpenAI | $2,048 |
| Claude SonnetAnthropic | $2,727 |
| Claude OpusAnthropic | $13,635 |
cheapest · public list prices as of 2026-06 · planning estimate, not a quote
what_drives_it
Where the money goes
Volume with short turns: a cheap model handles it well, and the single tool call adds only a modest second turn.
The cheapest option here, Gemini 1.5 Flash, comes to about $61.43 a month against $123 on GPT-4o mini. Whether the cheaper model fits is a question for your evaluation set, not the price sheet. The bigger lever is usually the workload itself: caching re-sent context, trimming what each turn carries, and capping the tool loop move the bill more than swapping models does.
faq
Questions & answers
- How much does a email triage agent cost per month?
- On GPT-4o mini, about $123 a month at 6,000 requests a day with the assumptions below. The cheapest model compared here, Gemini 1.5 Flash, runs about $61.43 for the same workload. Your real figure moves with volume and tokens, so tune it in the calculator.
- What makes a email triage agent expensive?
- Volume with short turns: a cheap model handles it well, and the single tool call adds only a modest second turn.
- Which model is cheapest for a email triage agent?
- Gemini 1.5 Flash, at about $61.43 a month for this workload. Cheaper is not automatically better: a model that needs retries or longer prompts can cost more in practice, so test the candidates on your own evaluation set before committing.
A cost estimate is a start. Making an agent cheap in production is the work.
Prompt caching, context trimming, and the right model per step usually cut an agent's bill by more than half. Book a call, or leave your email and I'll reach out.
Prefer proof first? See how this plays out in real case studies →