How many tokens does a 3-point story take to build with AI?
A 3-point user story typically takes about 528,000 tokens to build with an AI coding agent (roughly 212,000–1,160,000 depending on complexity), about $1.53 at today's GPT-4o prices ($0.6200–$3.35). This is the build-time (developer) cost, not the cost of running the finished feature.
A 3-point story is the workhorse estimate on many boards — a task with a clear shape but several files and a real test cycle. The agent iterates more, so input tokens climb into the hundreds of thousands.
Token & cost ranges
Cost uses GPT-4o at current prices. Verified 29 July 2026.
| Scenario | Input tokens | Output tokens | Total tokens | Est. cost |
|---|---|---|---|---|
| Low (simple) | 200,000 | 12,000 | 212,000 | $0.6200 |
| Typical | 500,000 | 28,000 | 528,000 | $1.53 |
| High (complex) | 1,100,000 | 60,000 | 1,160,000 | $3.35 |
Typical cost by model
The same typical token estimate priced across a few common coding models.
| Model | Typical cost | Range (low–high) |
|---|---|---|
| GPT-4o | $1.53 | $0.6200 – $3.35 |
| GPT-4o mini | $0.0918 | $0.0372 – $0.2010 |
| Claude Sonnet 4.6 | $1.92 | $0.7800 – $4.20 |
| Gemini 2.0 Flash | $0.0612 | $0.0248 – $0.1340 |
How we estimate this
A 3-point story is a normal half-day-to-day task touching several files with a real test pass and some back-and-forth. Input tokens dominate because an agent re-sends file context on every turn; output is the generated code, diffs and tests. Treat these as planning ranges, not a quote — your real usage depends on codebase size, model, prompt caching and how many iterations it takes.
Examples of this size
- A new endpoint with validation and tests
- A feature toggle across UI and API
- Integrating a small third-party SDK
Estimate it with your own model
Plug in your model, complexity and prompt-caching to get a tailored range.
Frequently asked questions
How many tokens does a 3-point user story take to build with AI?+
A 3-point user story typically takes about 528,000 tokens to build with an AI coding agent (roughly 212,000–1,160,000 depending on complexity), about $1.53 at today's GPT-4o prices ($0.6200–$3.35). This is the build-time (developer) cost, not the cost of running the finished feature.
Is this the cost to build it or to run it?+
This is the build-time (developer) cost — the tokens an AI coding agent burns while implementing the work. Running the finished feature is a separate, usually much smaller per-request cost.
Why is the range so wide?+
A 3-point story is a normal half-day-to-day task touching several files with a real test pass and some back-and-forth. Input tokens dominate because an agent re-sends file context on every turn; output is the generated code, diffs and tests. Treat these as planning ranges, not a quote — your real usage depends on codebase size, model, prompt caching and how many iterations it takes.
How can I lower the cost?+
Split large work into smaller pieces, enable prompt caching so re-sent context is billed at the cheaper cache-read rate, and use a cheaper model for routine steps. Try your own model in the Cost to build X estimator.