Wondy.
·AI Tools·27 min read·Wondy

ChatGPT Prompt for Roth vs Traditional 401(k), Run on a $92,000 Salary

A ChatGPT prompt to choose Roth or traditional 401(k) in 2026, run on a $92,000 salary. The break-even retirement rate is 19.3%, not 22%.

The choice arrives as two radio buttons in a benefits portal. Pre-tax. Roth. Underneath them sits a percentage box and a Save button, and nothing on the screen tells you which one is about to cost you money.

So people paste the question into a chat window, which is a better move than it sounds and worse than it looks. The difference between the two buttons is only about timing. A traditional contribution comes out before tax, cuts this year's taxable income, and the government collects when you withdraw. A Roth contribution comes out of money you have already been taxed on, and the withdrawal is not taxed at all. Everything else follows from one comparison: your tax rate now against your tax rate then.

The trouble is that the most repeated answer to that comparison was built on a law with an expiry date, and the expiry date was deleted. Call it the sunset premise. It ran through nearly every Roth article published between 2018 and 2024, it is thick in the training data, and it stopped being true on July 4, 2025.

Why "rates go up in 2026" is dead law

The 2026 sunset never happened. The One Big Beautiful Bill Act made the current rate schedule permanent in July 2025, which pulls the load-bearing beam out of the standard Roth argument.

IRS Rev. Proc. 2025-32 puts it in one sentence at section 2.01:

"Section 70101 of the OBBBA amends 1(j) to make the tax rate tables that were effective for taxable years beginning after December 31, 2017, and before January 1, 2026, permanent. The existing seven tax rates of 10%, 12%, 22%, 24%, 32%, 35%, and 37% remain in effect for individual taxpayers."

OBBBA is Public Law 119-21, signed July 4, 2025. Before it, the individual rates were legislated to revert at the end of 2025, and the top rate was set to climb back to 39.6%. That scheduled increase was the whole engine of the advice: pay tax now while it is cheap, because it will not be cheap for long. The rest of the pre-2018 bracket schedule gets quoted freely in this argument, and my primary sources support only the top-rate figure, so the top rate is all I will state.

Permanent here means until Congress changes it, which is a status rather than a promise. A status you have to argue against is still very different from a date already on the calendar. The same law rewrote student loan repayment, and the RAP plan change check covers that side of it. For 2026 the deferral limit is $24,500 and the IRA limit is $7,500, and the mid-year 401(k) checkup works through the match, fee and rebalancing dials rather than the tax toggle.

The prompt, and the numbers I put in it

One profile, one cold run, no web access. The prompt hands the model every input it needs, asks it to show the break-even arithmetic, and ends by asking what it would need to know that I never told it.

I ran this on Claude Opus 5 on August 18, 2026, rather than in a ChatGPT window. The prompt is not model-specific and behaves the same in either, and the current free tier handles this length of question without a subscription. The profile: single, 35, no dependents, $92,000 salary, standard deduction, no state income tax, $12,000 a year into the plan, an employer match of 4% of salary at $3,680, 30 years to retirement, 6% nominal return, 2.5% inflation, and $24,000 a year of Social Security in 2026 dollars.

AI 답변 (실제 실행)Claude (Opus 5)2026-08-18 실행

Prompt: I'm 35, single, no dependents. My 2026 salary is $92,000, I take the standard deduction, and I live in a state with no income tax. I put $12,000 a year into my 401(k) and my employer matches 4% of salary ($3,680). I have 30 years to retirement. Assume a 6% nominal return and 2.5% inflation. In retirement I expect about $24,000 a year from Social Security (in 2026 dollars) plus 401(k) withdrawals.

Should the $12,000 go into the Roth 401(k) or the traditional 401(k)? Please: 1) Tell me my current federal marginal tax bracket under 2026 law and how much tax the traditional contribution saves me this year. 2) Compute the break-even retirement tax rate for a single year's $12,000 contribution held 30 years, showing the math. 3) Estimate what my effective federal tax rate on withdrawals would likely be in retirement given this profile. 4) List what you would need to know that I haven't told you. 5) Note anything in federal tax law enacted after 2025 that changes the usual "rates go up in 2026, so choose Roth" advice, and flag anything you're not sure about.

(excerpt: the framing note, both break-even calculations, the retirement estimate, the post-2025 law table row on catch-ups, and the closing warning. The full run also carried a second under-saving scenario and thirteen missing inputs.)

One framing note before the math. The usual advice you've probably read — "the TCJA rates expire at the end of 2025, so rates go up in 2026, so lock in Roth now" — is obsolete. That expiration didn't happen. More on that in section 5, but it changes the starting point: your bracket is 22%, not the 25% it would have been.

Both land inside the 22% bracket ($50,400–$105,700), so the entire $12,000 deduction is worth 22% — no bracket-straddling to complicate it.

Marginal bracket: 22%. Traditional saves you $2,640 in federal tax this year.

Notice what dropped out: the 6% return and the 30 years cancelled completely. That's not a coincidence and it's the single most useful fact in this whole decision. Both accounts grow by the same multiplier, so the multiplier appears on both sides and vanishes. Roth vs. traditional is not a bet on compounding, or on how long you have, or on how well you invest. It is purely a bet on your marginal rate today vs. your marginal rate when the money comes out.

Break-even = 22%. If your retirement rate on these dollars is below 22%, traditional wins. Above 22%, Roth wins. Exactly 22%, it's a coin flip.

The alternative framing (and why I don't think it's yours). A lot of write-ups instead assume you contribute $12,000 either way and invest the $2,640 tax saving in a taxable brokerage account. That version penalizes traditional, because the side account leaks tax every year on dividends and rebalancing, and again on capital gains at the end:

  • $2,640 growing at ~5.5% net of drag for 30 years: $2,640 × (1.055)^30 = $2,640 × 4.984 = $13,158
  • Capital gains tax at 15% on the $10,518 gain: −$1,578 → $11,580 net
  • Break-even becomes: $68,922 × (1 − t) + $11,580 = $68,922 → t = 16.8%

With a more tax-efficient fund (say 0.3% drag) it's about 17.7%. So that framing gives a break-even around 17–18%, which is a materially tougher bar for traditional.

Which one applies to you? The 22% one. The side-account framing only binds if you're contribution-limited — if you're already at the annual deferral cap and the only way to shelter more is to make your contributions denser by going Roth. You're at $12,000 against a 2026 elective deferral limit of about $24,500. You have plenty of headroom.

In Scenario B you've saved enough that you've pushed past the torpedo: 85% of your Social Security is already taxable, the multiplier is exhausted, and your marginal rate falls back to a plain 12%.

  • Scenario B (my base case): 12% marginal vs. 22% today → traditional wins by 10 points. Clear.

The One Big Beautiful Bill Act (P.L. 119-21, enacted July 4, 2025) made the TCJA individual rate structure permanent. The scheduled 2026 sunset — which would have restored the 15 / 25 / 28 / 33 / 35 / 39.6% brackets — did not happen. For most of 2017 through 2025, "rates are legislated to rise on January 1, 2026, so buy Roth at today's discount" was a legitimate, statute-backed argument. As of 2026 it is simply wrong, and a lot of blog posts and calculators still haven't caught up.

(from the post-2025 changes table) SECURE 2.0 §603: mandatory Roth catch-up — Required for catch-up contributions if prior-year FICA wages exceed ~$145,000 (indexed); in effect from 2026. Doesn't bind you now. Will if your income grows and you're 50+.

Do not miss this, because it's where the traditional case actually breaks: the entire argument assumes you invest the $2,640, ideally by raising your deferral toward $15,385. If the $2,640 goes to spending, you have not chosen traditional over Roth — you have chosen to save less. Roth's real-world edge over traditional is largely behavioral, and it's not a trivial edge.

What held up when I recomputed all of it

Nineteen numbers and claims, checked line by line against the IRS releases and rerun in a calculator. Fifteen held exactly, one was a superseded figure, and three rest on things my sources do not support.

CheckedResult
All seven 2026 single brackets, $16,100 standard deductionMatches Rev. Proc. 2025-32 exactly
Taxable income $63,900 traditional against $75,900 Roth, both inside 22%Correct, and the no-straddle point is why the rest stays clean
Federal tax $8,770 against $11,410, difference $2,640Correct to the dollar
Growth factor 1.06 to the 30th at 5.7435, pre-tax equivalent $15,385Correct
Equal-cost break-even of 22%, growth factor cancelling outCorrect, and the sharpest observation in the run
2026 deferral limit of about $24,500Correct, and it hedged on a number it had right
Social Security thresholds $25,000 and $34,000, never indexedCorrect, and it built the retirement estimate around that
Senior deduction of $6,000, phasing out above $75,000, 2025 through 2028 onlyCorrect, and it warned me not to project it into 2056
Retirement estimate of 12.1% effective on withdrawalsMatches my own run to one decimal
Roth catch-up wage line at "~$145,000"Wrong. Notice 2025-67 raised it to $150,000
"In effect from 2026" stated flatlyOverstated. The rule text applies after 2026, covered below
Pre-sunset brackets "15 / 25 / 28 / 33 / 35 / 39.6%" and "the 25% bracket"Unsupported by my sources, which carry only the 39.6% top rate
The 16.8% break-even, built on a 5.5% "net of drag" returnArithmetic correct, drag assumption supplied by the model

The pattern is not the one I set out to catch. Every figure sitting in an IRS release the model had already seen came back right, including two I expected it to fumble. The one it got wrong is a figure that was recently replaced by a newer one, and the two it overreached on are the places where it filled a gap with something plausible instead of leaving the gap. That is the same failure shape as an AI pass over a 10-K: fluent where the source exists, inventive where it does not.

The break-even is 19.3%, and the model argued for 22%

Both numbers are right. They answer different questions, and the gap between them is the most interesting thing in this exercise.

Start with the current rate, because everything hangs on it. Take $92,000, subtract a $12,000 traditional contribution and the $16,100 standard deduction, and taxable income is $63,900. Choose Roth instead and it is $75,900. Both sit inside the 22% band, which runs from $50,400 to $105,700, so the entire contribution is deducted at one rate with no bracket straddling. That makes the marginal tax rate exactly 22% and the first-year deduction worth $2,640. If that arithmetic feels unfamiliar, the W-4 audit prompt walks through where marginal rates and withholding actually meet.

Now the fork. Follow one $12,000 contribution for 30 years at 6%, a growth factor of 5.7435.

One $12,000 contribution, 30 years, 6% nominal
Roth 401(k)
  • $12,000 in, $68,922 out
  • No tax on the withdrawal
  • No RMD while you are alive
Headline number
Traditional, saving invested
  • Same $68,922, taxed at rate r
  • Plus $2,640 grown to $15,163
  • Less 15% on the gain: $13,284 net
Traditional, saving spent
  • $68,922 taxed at r, nothing beside it
  • Roth wins at any rate above zero
  • Not a tax choice, a savings cut
Growth factor 1.06 to the 30th power is 5.7435. Long-term capital gains assumed at 15%. Set the middle column equal to the left one and 68,922r = 13,284, so r = 19.3%.

Set the middle path equal to the Roth path and the equation is 68,922 times r equals 13,284. The break-even retirement rate is 19.3%, not the 22% you pay now, because the side account owes capital gains tax that the 401(k) never owes.

The model rejected that framing, and its reasoning is good. If you are nowhere near the $24,500 limit, it argued, you would not open a side account at all. You would raise your deferral from $12,000 to $15,385, keep your take-home pay exactly where it was, and then the growth factor cancels on both sides and the break-even is a flat 22%.

I lean the other way, on one observation about the form. Enrollment screens take a percentage, and someone who flips the toggle from pre-tax to Roth almost never reaches over and raises that percentage to compensate. The 22% answer describes a person who re-optimizes the whole contribution the same afternoon. The 19.3% answer describes a person who clicked a radio button. Most of the traffic on this question is the second person, so 19.3% is the number I would put on the wall, while conceding that anyone who does raise the deferral has both the cleaner deal and the better math.

Single filer, $92,000, 2026 brackets from IRS Rev. Proc. 2025-32. Below the break-even line, traditional wins. The retirement rate is the effective federal rate on 401(k) withdrawals in the projection below.Marginal rate today 22%, Break-even (saving invested) 19.3%, Projected retirement rate 12.1%Marginal rate today22%Break-even (saving inve…19.3%Projected retirement rate12.1%
Single filer, $92,000, 2026 brackets from IRS Rev. Proc. 2025-32. Below the break-even line, traditional wins. The retirement rate is the effective federal rate on 401(k) withdrawals in the projection below.

What your retirement rate will actually be

12.1% on this profile, which is seven points under the break-even and ten under today's bracket.

The balance first. At a real return of 3.5%, thirty years of a $12,000 deferral compounds to $619,472 in 2026 dollars, and the $3,680 employer match adds $189,971, for a total of $809,444. A 4% withdrawal is $32,378 a year. That match figure matters for a reason unrelated to its size: unless your plan offers a Roth match and you elect it, employer money lands in the pre-tax bucket no matter which button you chose, so you end up holding a traditional balance either way.

Then the tax. Add half of $24,000 in Social Security to the $32,378 withdrawal and provisional income is $44,378, above the $34,000 line, so up to 85% of the benefit becomes taxable, or $20,400. That puts AGI at $52,778. Subtract the $16,100 standard deduction and the $2,050 age-65 addition and taxable income is $34,628, sitting in the 12% band. Tax owed is $3,907, which is 12.1% of the withdrawal.

The model ran the same projection in 2056 nominal dollars rather than today's, on the grounds that the Social Security thresholds of $25,000 and $34,000 have never been indexed while brackets and the standard deduction are. It landed on 12.1% too. Its method is the more careful one, and it matters for the same reason reading a CPI report properly changes what a number means: two figures in one formula can move at different speeds. The 2.5% inflation assumption underneath all of this is a guess, and your own inflation rate is not the published one.

Do catch-up contributions have to be Roth in 2026?

Ask your plan administrator, because 2026 is the year where the correct answer is genuinely plan-dependent, and nearly every page ranking for this question picks a side anyway.

Three IRS documents point slightly different ways. Notice 2025-67 says the 2025 wage figure is "used to determine whether an individual's catch-up contributions to an applicable employer plan ... for 2026 must be designated as Roth contributions, is increased from $145,000 to $150,000." The final regulations, announced in IR-2025-91 on September 15, 2025, "generally apply to contributions in taxable years beginning after Dec. 31, 2026," with earlier years run on "a reasonable, good faith interpretation of statutory provisions." That same release says the final rules "do not extend or modify" the transition relief in Notice 2023-62, which "generally ends on Dec. 31, 2025."

Read together: the grace period is over, the regulation text does not bite until 2027, and 2026 sits in the good-faith window where your plan's own reading governs. This is also the one place the run went stale, quoting the old $145,000 line and calling the rule flatly in effect for 2026. The catch-up amounts themselves are $8,000 from age 50 and $11,250 for ages 60 through 63.

Where the answer flips

Four things, ordered by how much each one moves the outcome.

Behavior is the big one, and it is the thing no model can see. The traditional case is built on the $2,640 going somewhere productive. Spend it and you have not made a tax decision at all. The model said this more bluntly than I would have, which is the only place in the run where it sounded like a person.

A second income in retirement comes next. A spouse's pension, an inherited IRA, a large conversion, or a bigger Social Security benefit than the $24,000 I assumed all stack under the withdrawal and push it toward the 22% band. There is also a squeeze inside the Social Security formula itself, where an extra dollar of withdrawal drags additional benefit dollars into taxable income and lifts the effective rate above the printed bracket. How big that effect is on this profile is something I did not compute, and the missing input is exactly where your provisional income lands.

Maxing out flips the frame entirely. A Roth dollar is denser than a traditional dollar, because $12,000 of Roth money costs $15,385 of salary. Fill the $24,500 limit with Roth and you have sheltered the pre-tax equivalent of $31,410, which no traditional contribution can match at the same cap. For anyone at the limit, the break-even calculation understates Roth.

Everything left out leans the same direction. State income tax appears nowhere above, and where you live at 65 may not be where you live now. Medicare premium surcharges, the tax treatment your heirs will face, and the absence of required minimum distributions on Roth accounts all sit outside the equation and all favor Roth. The IRS puts the RMD point plainly: the rules "do not apply to Roth IRAs or Designated Roth accounts while the owner is alive."

Run it on your own numbers

Five steps, and the only one that needs care is the third.

From your pay stub to a decision
  1. 01
    Find your marginal rate

    Salary minus contribution minus standard deduction, then read the 2026 bracket table.

  2. 02
    Multiply for this year

    Contribution times that rate is what the traditional deduction hands back.

  3. 03
    Decide what happens to it

    Invested, spent, or added to the deferral. This choice sets your break-even.

  4. 04
    Estimate the retirement rate

    Project the balance, take 4%, add Social Security, subtract the deduction.

  5. 05
    Compare, then check the plan

    Roth option, split contributions, Roth match. Only the plan document knows.

The prompt above does steps 1 through 4 in one pass. Recompute the arithmetic before acting on any of it, and read every figure your AI cites back to the IRS release it came from.

Six things stay outside the whole calculation and belong in your own margin. Permanent rates last until the next Congress decides otherwise, which is not a thirty-year guarantee. State tax is absent from every number above. Your plan document holds the Roth option, the split rule and the match rule. Your real Social Security benefit is on your SSA statement. The senior deduction of $6,000 expires after 2028 under current law, so anyone retiring in 2056 should not count it. And 6% nominal is a convention rather than a forecast, which the four recession numbers piece is a reasonable antidote to.

FAQ

Is a Roth or traditional 401(k) better for me in 2026?

It depends on one comparison and one behavior. The comparison is your marginal tax rate now against your marginal rate when you withdraw. On the profile I ran, a single filer earning $92,000 with $12,000 going into the plan, the current marginal rate is 22% and the projected retirement rate on those withdrawals is 12.1%, so traditional wins by a wide margin. The behavior is what happens to the $2,640 the deduction hands back. Invest it and traditional keeps its edge down to a retirement rate of 19.3%. Spend it and Roth wins at any positive retirement rate, because you did not choose a tax treatment, you chose to save less. Two facts move the answer toward Roth regardless: your employer match lands in a pre-tax bucket unless your plan offers a Roth match and you elect it, so you will hold a traditional balance either way, and Roth accounts carry no required minimum distributions while the owner is alive.

What retirement tax rate makes a traditional 401(k) beat a Roth?

19.3%, if you contribute the same dollar amount either way and put the tax saving in a taxable brokerage account. Here is the arithmetic on a single $12,000 contribution held 30 years at 6%. The growth factor is 1.06 to the 30th power, or 5.7435. A $12,000 Roth contribution becomes $68,922 and none of it is taxed. The traditional contribution becomes the same $68,922 and is taxed at your retirement rate. The $2,640 you saved on this year's taxes grows to $15,163 in a taxable account, and 15% capital gains tax on the $12,523 gain leaves $13,284. Set the two sides equal and 68,922 times r equals 13,284, so r is 19.3%. Below that rate traditional wins, above it Roth wins. The number is not your current bracket of 22%, because the side account pays tax that the 401(k) does not.

Did tax rates go up in 2026 when the 2017 tax cuts expired?

No, because they did not expire. The One Big Beautiful Bill Act, Public Law 119-21, was signed on July 4, 2025, and IRS Rev. Proc. 2025-32 describes the effect in section 2.01: "Section 70101 of the OBBBA amends 1(j) to make the tax rate tables that were effective for taxable years beginning after December 31, 2017, and before January 1, 2026, permanent. The existing seven tax rates of 10%, 12%, 22%, 24%, 32%, 35%, and 37% remain in effect for individual taxpayers." The 2026 brackets for a single filer run 10% up to $12,400, 12% to $50,400, 22% to $105,700 and 24% to $201,775, with a standard deduction of $16,100. This matters for the Roth decision because most published Roth advice from 2018 through 2024 was built on the sunset that was scheduled and then repealed. Permanent means until Congress changes it, which is a policy status rather than a promise.

Do my 401(k) catch-up contributions have to be Roth in 2026?

Ask your plan administrator, because 2026 is the year the answer is genuinely plan-dependent. Three IRS documents point in slightly different directions. Notice 2025-67 raises the wage line "from $145,000 to $150,000" and says the 2025 wage figure decides whether catch-ups "for 2026" must be designated as Roth. The final regulations, announced in IR-2025-91 on September 15, 2025, "generally apply to contributions in taxable years beginning after Dec. 31, 2026," and tell plans to use "a reasonable, good faith interpretation of statutory provisions" for earlier years. The same release says the final rules "do not extend or modify" the transition period from Notice 2023-62, which "generally ends on Dec. 31, 2025." So the grace period is over, the regulation text does not bite until 2027, and 2026 sits in the good-faith window where your plan decides how to operate. The catch-up amounts for 2026 are $8,000 at 50 and older and $11,250 at ages 60 through 63.

What can ChatGPT not know about my Roth vs traditional decision?

Whether you are the kind of person who invests a tax refund. That single unknown is worth more than every bracket table in the answer, because the entire case for traditional rests on the $2,640 going into an account rather than into your checking balance, and no model can observe that about you. Four more sit close behind. Your state income tax, both where you live now and where you will live at 65, is missing from every calculation above. Your plan document decides whether a Roth option exists at all, whether you can split contributions, and whether the employer match can be Roth. Your actual Social Security benefit is on your SSA statement, not in the $24,000 I assumed. And any second source of retirement income, a spouse's pension or an inherited IRA or a large conversion, pushes withdrawals into a higher band and moves the answer toward Roth. Give the model those five and its answer gets sharper. Withhold them and it will still answer, confidently.

Quick O/X quiz
  1. 01

    Congress let the 2017 tax cuts expire at the end of 2025, so federal income tax rates went up in 2026.

  2. 02

    If you contribute the same dollar amount either way and invest the tax saving in a taxable account, the break-even retirement tax rate is lower than your current bracket.

  3. 03

    Your employer match goes into the same tax bucket you pick for your own contribution.

For the accounts around this one, a four-dial checkup of the same 401(k) covers the match and fee levers the tax toggle never touches, and what an AI can and cannot see in a linked account is worth reading before you connect anything. For the same method pointed at other numbers, there is a payoff order for what you owe and a screener prompt for undervalued stocks.

Sources