What's happening
Not at a normal reset: Base44's Credits documentation says "any credits you have not used expire when they reset", and that applies to every published plan (Free, Starter, Builder, Pro and Elite). The one exception the docs describe is an upgrade from a monthly plan to a higher monthly plan, which carries unused credits over for one more cycle, up to one month of your plan's credits. Base44's Terms of Service (section 10.4) reserve the right to grant rollover to some customers at Base44's discretion; no plan includes it by default. Paid plans reset on the day of the month you subscribed, at your purchase time in UTC. Combined with two related facts (the documented way to get more mid-cycle is an upgrade, and the AI agent's burn is hard to predict), there is no right amount to buy: under-buy and you run dry mid-cycle, over-buy and the surplus expires. Fix it operationally by tracking the credit balance every Monday and computing weekly burn against days remaining, timing large agent-heavy builds to the first 1-2 weeks of the cycle, saving documentation and small visual tweaks for cycle-end, and matching plan tier to predicted usage instead of paying two tiers above just-in-case.
Your billing cycle just renewed. Your credit balance reset. The 200 unused credits you carefully avoided burning last month are gone. You paid for them, you did not use them, and they expired on your reset date. Next month you'll either over-use and run dry early, or under-use and lose more value at the next renewal.
The pattern shows up across base44 reviews. Combined with two related issues — excessive credit burn on minor changes and running out of credits mid-cycle — the credit system creates a no-win economic dynamic. Buy too few credits and the documented way out is an upgrade. Buy too many and the surplus expires. Time your work poorly and you both waste credits and run out before the cycle ends.
The frustration compounds because Base44's credit usage page shows what you have spent but does not project your burn rate. You discover the loss only on renewal day. By then it is too late to do anything about it.
When exactly do Base44 credits reset?
Paid plans reset monthly on the day of the month you subscribed, at the exact time of your original purchase in UTC. Subscribe on 5 July and credits reset on the 5th of every month, which can land on the 4th in your own timezone if you are behind UTC. The Free plan runs two clocks: up to 5 credits a day and 25 a month, with the daily allowance resetting at midnight in the workspace owner's timezone (UTC if none is recorded).
Source: docs.base44.com/Account-and-billing/Credits, read 17 October 2026. If your credits have not refreshed, the docs say to wait until your reset time has passed in UTC and check again.
Why this happens
Use-it-or-lose-it credit expiration is a written policy, not a bug. Base44 states it in two places:
- The Credits documentation: "Your credits refresh automatically on a set schedule, and any credits you have not used expire when they reset." The one exception is an upgrade from a monthly plan to a higher monthly plan, after which unused credits "carry over for one more cycle, up to one month of your plan's credits, then expire".
- The Terms of Service, section 10.4 (last updated June 22, 2026): AI credits "are currently granted on a recurring monthly basis and any unused credits do not roll over to the subsequent month". The same section reserves Base44's right, "in our sole discretion, to allow unused AI Credits (or any portion thereof) to roll over" for all or certain customers, and says unused credits also expire when a plan or account is cancelled.
Base44 does not say why it chose this model. The usual reasons subscription products expire credits (predictable monthly revenue, a nudge to keep building, simpler accounting when plans change mid-cycle) are our reading, not Base44's statement.
The reason base44's version of this model is unusually painful is the interaction with two other policies:
- The documented way to get more mid-cycle is an upgrade. Base44's docs describe upgrading or adjusting your plan, and describe no credit top-up — see running out of credits mid-cycle. Combined with no rollover, you cannot buffer between cycles.
- AI agent burn is hard to predict. Even careful users get hit by excessive-credit-burn-minor-changes: one user reported 15 credits for a single CTA link change, against Base44's own estimate of about 0.5 for a simple text change. Predicting your monthly need is genuinely hard.
The combination forces users into bad outcomes. If you under-buy, you run dry mid-cycle and have to upgrade or wait. If you over-buy, the excess expires. There is no "right" amount because the AI's burn rate is not deterministic.
Sources: docs.base44.com/Account-and-billing/Credits, read 17 October 2026, and base44.com/terms-of-service (section 10.4), read 23 September 2026; feedback.base44.com (multiple credit-system threads); G2 reviews discussing pricing dissatisfaction; Henry Collins on Medium documenting credit-burn anomalies; producthunt.com/products/base44/reviews on plan policies.
How to reproduce
This is platform policy, not a bug — there is no "reproducing" it. To verify the expiration rule on your account:
- Note your current credit balance and your plan's monthly allowance.
- Note your reset date and the time you originally subscribed.
- Use less than your full allowance for the cycle.
- After your reset time has passed in UTC, check your credit balance.
- The balance will equal your plan allowance for the new cycle, not (allowance + leftover from prior cycle). The exception is a cycle in which you upgraded from a monthly plan to a higher monthly plan: then the docs say unused credits carry over for one more cycle.
This confirms the no-rollover behavior on your specific account.
Step-by-step fix
There is no platform-side fix because this is policy, not bug. The fix is operational — manage your usage to minimize waste.
1. Track usage weekly
Note your credit balance every Monday morning. Subtract from last Monday's balance to get a weekly burn rate. Project the burn rate against days remaining in the cycle. If projected use exceeds your plan allowance, slow down. If it is below the allowance, accelerate non-urgent work to avoid waste.
A simple shared spreadsheet, shown here for a Pro plan's 500 message credits:
| Week | Mon Balance | Weekly Burn | Days Left | Projected Use by Reset |
|------|-------------|-------------|-----------|------------------------|
| W1 | 500 | - | 28 | - |
| W2 | 430 | 70 | 21 | 280 |
| W3 | 350 | 80 | 14 | 310 |
| W4 | 240 | 110 | 7 | 370 |
In this example the projection lands near 370 of 500, so about 130 credits are on course to expire: pull non-urgent agent work forward.
2. Time large builds to early in the cycle
Run AI-agent-heavy work — new features, large refactors, schema changes — in the first 1–2 weeks after renewal. This buys you time to course-correct if a regression loop or hallucination burst eats your budget unexpectedly.
3. Save low-effort work for cycle-end
Documentation, content edits, small visual tweaks — work that costs few credits — fits naturally late in the cycle when you want to use up remaining balance without risking overrun.
4. Match plan tier to predicted usage
Two patterns work:
- Build months on higher tier, maintenance months on lower tier. Upgrade before known heavy work; downgrade for quiet stretches.
- Stay on the smallest plan that covers your typical month. Accept occasional cycle-end runouts rather than persistent waste.
The wrong pattern is paying for a plan two tiers above your typical usage just to avoid running out: most of the difference between the two allowances expires every month. Before you choose, see what each Base44 plan costs per month at your edit pace and integrations.
5. Avoid mid-cycle plan upgrades for buffer
Upgrading mid-cycle takes effect immediately after you pay. Base44's docs say a move from a monthly plan to a higher monthly plan charges "the full price difference between the 2 plans" and carries unused credits over for one more cycle, while any upgrade that involves a yearly plan is prorated and unused credits "expire at your next reset as usual"; the upgrade window shows which applies before you confirm. The catch is the other direction: a downgrade only takes effect at the end of your current billing period, so the higher plan keeps renewing until you step back down. Unless your usage will sustain the higher tier, the upgrade trades a small short-term gain for a structurally larger monthly cost.
6. Track credit-burn anomalies
If a single message's Credits Used figure is above Base44's ~3–4 estimate for an app-wide change, log it. Patterns matter: certain prompts and certain code patterns burn predictably more. Build internal docs of what to avoid. Long-context regressions and hallucinated fields are common offenders — see related fix pages.
DIY vs hire decision
This is fully DIY. The skills are clerical: track usage, plan timing, choose the right plan tier. Most teams build a small shared spreadsheet and check it weekly. There is no specialist work to outsource.
There is a larger question sitting behind the clerical one. If credit timing is something you now manage on a spreadsheet, it is worth asking whether the plan is still worth renewing at all — our full Base44 review prices the keep, fix and leave options against each other rather than treating renewal as the default. If the answer is no, what cancelling a Base44 plan does to your apps and credits covers the step before you stop paying: unused credits expire on cancellation, and the workspace drops to the Free plan at the end of the period.
Hire only if you suspect deeper issues that are inflating your burn rate beyond normal — the excessive-credit-burn-minor-changes pattern, AI agent regression loops, or context-window-driven rework. In those cases the credit waste is a symptom of a fixable architecture issue. We diagnose burn-rate anomalies as part of the $497 audit and identify the underlying causes.
Need this fix shipped this week?
If your monthly base44 spend is meaningful and you suspect waste, the diagnostic audit is the right product. We analyze a month of your usage data, identify burn-rate anomalies, recommend plan-tier adjustments, and flag any underlying technical issues inflating the rate.
Order a $497 audit — fastest path to understanding where your credits actually go.
Related problems
- Excessive credit burn on minor changes — the real source of monthly overruns.
- Running out of credits mid-cycle — the other half of the no-win credit dynamic.
- Vendor lock-in via SDK dependency — when credit waste persists, migration math starts to favor leaving.
- What running a Base44 app really costs each month — expiring credits are one of the hidden line items that inflate your effective monthly cost.