Claude-Built FIRE Calculator: Cost & Risks
A Show HN FIRE calculator was built with Claude. See what runs locally, which AI costs are undisclosed, and the assumptions to verify.
By AI Pricing Guru Editorial Team
AI Pricing Guru articles are maintained by the editorial workflow behind the site: daily pricing snapshots, provider source checks, and review passes for model launches, subscription limits, and billing changes.
TL;DR
- A new Show HN project uses a day-by-day Python simulation to estimate when a user can coast or retire; its author says Claude wrote the code, tests, and design log.
- The finished calculator runs locally without an AI API. Claude is optional for editing plans, so recurring model cost can be zero after setup.
- The creator did not disclose the Claude model, route, token usage, subscription tier, or build cost. The live Claude tables below are current options—not a reconstruction of this project’s bill.
- Treat every result as a scenario, not financial advice: the model is deterministic, has no sequence-of-returns risk, and has not been reviewed by a financial professional.
Current Claude plans for an optional Claude Code workflow
| Plan | Current price | Tier | Published limits and access |
|---|---|---|---|
| Claude Free | $0 | free | Limited Sonnet access, resets every 5 hours |
| Claude Pro | $20/mo | pro | Approx 5x Free, priority during peak, extended context. Claude Fable 5 access continues via usage credits beginning July 20, 2026, with a one-time $100 Fable credit. |
| Claude Max (5x) | $100/mo | power | Claude Fable 5 is included beginning July 20, 2026 at 50% of this plan's usage limits. Standard Max 5x usage remains about 5x Pro across the rest of the model picker. |
| Claude Max (20x) | $200/mo | max | Claude Fable 5 is included beginning July 20, 2026 at 50% of this plan's usage limits. Official Claude pricing page still lists Max 20x; price is retained from the prior verified source when the raw page only exposes Max pricing as 'From $100'. |
Checked September 27, 2026. Prices and plan notes are rendered at build time from the maintained subscription dataset.
Illustrative Claude API cost for the same token workload
USD per 1M tokens. Input and output rates are charted separately.
Estimate a separate Claude-assisted planning session
Assumes 75% input tokens and 25% output tokens using current per-million rates.
Claude Haiku 4.5
anthropic
$20.00
- Input share
- $7.50
- Output share
- $12.50
Claude Sonnet 5
anthropic
$40.00
- Input share
- $15.00
- Output share
- $25.00
Claude Opus 5.5
anthropic
$80.00
- Input share
- $30.00
- Output share
- $50.00
Current Claude API rates—the project's build route was not disclosed
| Model | Provider | Input / 1M | Cached / 1M | Output / 1M |
|---|---|---|---|---|
| Claude Haiku 4.5 | anthropic | $1.00 | $0.1 | $5.00 |
| Claude Sonnet 5 | anthropic | $2.00 | $0.2 | $10.00 |
| Claude Opus 5.5 | anthropic | $4.00 | $0.2 | $20.00 |
Built from pricing.json at publish time.
A Claude-built FIRE calculator reached Show HN on September 24. The public Python project asks how long someone must remain at a high salary before coasting or retiring, then solves three targets with a day-by-day financial simulation.
Running the finished calculator does not require Claude or another model. AI cost applies only to the undisclosed development session or optional Claude Code use when changing plans.
What launched on Show HN
The creator says the tool was built “in a couple hours” and describes it as vibe-coded: the user specified the goal while Claude produced the code, tests, and design log. The repository records a long sequence of requested changes and decisions attributed to Claude, including a second-model design review.
The calculator simulates wages, taxes, expenses, investments, withdrawals, and capital-gains tax through age 112. It solves for the shortest high-salary work period under three goals: spend the portfolio down by the end, coast with core expenses covered until 60, or keep the portfolio from shrinking after 100.
This is a public source repository, but GitHub detected no license and the tree contained no license file when checked. Public visibility does not automatically grant permission to reuse or redistribute the code.
What Claude cost—and what remains free
The source does not identify the Claude model or whether development used a free chat, paid subscription, Claude Code allowance, or metered API. It also provides no token log, session transcript, invoice, elapsed AI time, or human-review time. A credible build-cost total cannot be calculated from the public evidence.
Anthropic’s official U.S. pricing page currently lists Claude Pro at $20 per month and Max from $100 per month; Claude Code is included with paid plans and shares their usage pool. Console-authenticated Claude Code follows standard API billing. Current direct rates are $1 input / $5 output per million tokens for Haiku 4.5, $2 / $10 for Sonnet 5, and $4 / $20 for Opus 5.5. These are current options, not a reconstruction of the author’s bill.
| Cost layer | What the source establishes | Buyer implication |
|---|---|---|
| Original AI build | Claude wrote the implementation, tests, and design log | One-time cost is undisclosed |
| Calculator runtime | Local Python with rich and plotext dependencies | No model call is required |
| Custom plan editing | Plain Python files; Claude Code is suggested as an interface | AI usage is optional and follows the user’s route |
| Financial validation | No professional review was performed | Expert review and user verification are separate costs |
The live subscription and API sections above show current Claude routes, not the route used by the author. See our full Anthropic pricing page, compare an alternative on OpenAI pricing, and test a known API token mix in the AI token calculator.
What the calculator does well
The project exposes assumptions as Python rather than hiding them behind a form. Users can model salary stages, a partner’s income, city moves, tax treatments, one-off expenses, and tax-advantaged accounts. Grid sweeps compare combinations, while what-if runs estimate how extra savings change the stopping date.
The repository includes tests and a decision log. These help reviewers inspect AI-generated design choices and regressions, but do not validate one household’s financial assumptions.
Where a retirement date can mislead
The model uses deterministic real returns and does not run Monte Carlo paths or sequence-of-returns scenarios. It omits Social Security by default and simplifies areas including required minimum distributions, early retirement-account access, employer matching, health savings accounts, tax deductions, and partner accounts.
Its tax inputs also mix year vintages and must be verified. The author explicitly warns that expense figures are rough examples and that the safety factor is a major lever. A precise stop date can therefore create false confidence when returns, inflation, healthcare, tax law, or personal spending diverge from the assumptions.
Who benefits—and who should wait
Technical users who want inspectable local scenarios gain the most. Comparing plans side by side helps reveal which assumptions drive the result while personal files remain gitignored.
People seeking a turnkey web calculator, audited tax treatment, probabilistic retirement analysis, or regulated advice should wait or use a professionally reviewed product. Non-programmers must also be comfortable installing uv, editing Python, and protecting sensitive financial inputs.
Developers evaluating the build method should compare cost per accepted feature, including prompt time, code review, tests, corrections, and maintenance. Our best AI for coding guide provides a broader baseline. A separately metered control is the Z.ai coding plan.
Affiliate disclosure: AI Pricing Guru may earn a commission from the sponsored Z.ai link at no extra cost to you. It does not affect this analysis.
Evidence boundary and practical advice
This launch cannot enter the AI Pricing Guru Labs leaderboard. The project is an application, not a callable model, and the author did not publish the exact model, route, prompts, token usage, or bill needed to reproduce its development economics.
Before relying on an output:
- Copy a shipped plan and replace every income, expense, tax, account, and life-event assumption.
- Run conservative and optimistic return cases rather than accepting one deterministic date.
- Compare account-enabled and taxable-only scenarios, then inspect cash-flow and withdrawal charts.
- Verify current tax rules and retirement-account limits against primary sources.
- Have a qualified professional review any decision that changes employment, insurance, or withdrawals.
Bottom line
The Show HN project demonstrates that an AI-assisted build can produce a substantial, tested local planning tool quickly. It does not establish how much Claude cost, that vibe-coded financial software is safe, or that the calculated retirement date is reliable.
Use it as an inspectable scenario engine. Keep AI optional, document the model route if you extend it, and treat the result as a prompt for deeper validation—not permission to quit a job.
Sources: the official budget-tools/fire repository, pinned source revision 5c10eef, its design log, the Hacker News discussion, Anthropic’s official Claude plan pricing, Claude Code page, and Claude API pricing documentation. Source, licensing, model-disclosure, and pricing claims checked September 24, 2026 between 05:24 and 05:40 UTC.