✓ Last verified: 2026-07-15· Source: official provider pricing page· Auto-monitored — report change →
An agent doesn't answer in one call — it loops: think, call a tool, read the result, think again. Each loop is a billed request, so cost multiplies by the number of steps. Enter your usage to see the real monthly bill and the cheapest model for the job.
The trap with agents: a task that looks like one question is really 5–20 LLM calls. A chatbot replies once. An agent reasons, calls tools, and re-reads its own growing context every step — so the same model that costs a fraction of a cent per chat message can cost 10–50× that per agent task. This calculator multiplies it out properly.
—
per month
—per task
—per year
—LLM calls / month
Cheapest models for this agent
Same workload, every model ranked by monthly cost.
Model
Cost / month
Per task
⚠️ Estimate using reference prices (July 2026) and list rates. Real agent cost depends heavily on how much context you resend each step — agents that re-feed the full history grow input tokens every loop. Type your real per-step token counts for the closest figure. · Report outdated price →
How agent cost actually adds up
The formula is tasks × steps × (input tokens × input price + output tokens × output price). The multiplier people forget is steps: a research agent might take 12 loops, a coding agent 20+. And because most agents resend the accumulated conversation each step, input tokens per step climb as the task goes on — the number you enter above is the average across the whole loop. Output is usually small per step (a short thought or a tool call), which is why input dominates the agent bill — and why prompt caching and context trimming matter more here than anywhere else.
How to cut it
Four levers, in order of impact: (1) model routing — use a small model for routine steps and a frontier model only for the hard ones; (2) step cap — stop runaway loops; (3) context trimming — don't resend the entire history every step; (4) prompt caching for the static system prompt and tool definitions. See the prompt caching guide and the caching savings calculator. Building a chatbot instead of an agent? Use the chatbot cost calculator. Whole app? The AI app cost estimator. Running more than one agent on the same task — fan-out, debate, supervisor or swarm patterns? This calculator prices a single agent; use the AI agent swarm cost calculator to see the multiplier those orchestration patterns add on top.
The AI Agent Cost Calculator estimates the monthly LLM spend of running an AI agent. Because an agent completes each task through several LLM steps — reasoning turns plus tool calls — the cost of one task is the per-step token cost multiplied across every step, then multiplied again by your monthly task volume. The main drivers are your tasks per month, the number of LLM steps per task, the average input and output tokens per step, the model you pick (which sets the per-token price), and any optimisation such as caching or cheaper models. Small per-step numbers compound quickly across steps and tasks.
The key trade-off to watch is step count versus model quality. A cheaper model may need more reasoning steps or retries to finish a task, so a lower per-token price does not always mean a lower total. Estimate cost across the full task, not a single call, and confirm the token counts you enter reflect real prompts — long system instructions and tool outputs are counted as input on every step.
Frequently asked questions
Why does an AI agent cost more than a single chatbot call?
An agent solves a task over many LLM calls — it reasons, calls a tool, reads the result, reasons again, and repeats. Each step is a full billed request, and the running context usually grows every step, so a single task can cost 5-20x a one-shot chatbot reply.
How do I cut AI agent costs?
Use a cheaper model for routine steps and reserve a frontier model for hard ones, cap the number of steps per task, trim the context you resend each step, and turn on prompt caching for the static system prompt and tool definitions that repeat on every call.