When free AI coding tiers reset, and what carries over

Short answer

One of the six free coding tiers that include model access publishes its reset to the second. GitHub: "Unused credits are forfeited, and your allowance resets to the full monthly amount at 00:00:00 UTC on the first day of each calendar month." Four of the other five publish no reset window at all.

The cadence sets the longest outage the tier can hand you: a monthly bucket is a bigger allowance and a three-week blackout if you empty it in week one. And "reset" describes two different arrangements, one of which takes something away from you on the way past.

Updated 14 November 2026. Clixad changed how its free credits work after this post was written. A new account no longer gets the 25,000-credit signup bonus that figures below rely on. Instead every budget model is free up to the daily allowance, seven models where this post may count five, and our own surveys and the daily streak add to that allowance for the day and the next rather than paying credits into the balance. The rest of the post stands as it was written.

The six windows

These are the same six tools as the shortlist that includes inference, because a tier you point your own API key at has no reset to speak of. Every cell is what the vendor publishes today, or says that they publish nothing.

What each free tier grants, when it comes back, and whether anything survives the boundary. Vendor figures, not engine output.
Free tier Reset window Published? Carries over?
GitHub Copilot Free 00:00:00 UTC on the first of each calendar month Yes, exactly No, forfeited
Cursor (Hobby) No quota published, so no window either No Unstated
Gemini CLI free tier A day, with no time zone or boundary given No Unstated
FreeBuff Six one-hour sessions "per day", boundary unstated No Unstated
Amp Free A daily allowance, reported in May 2026 as being paused or reduced for more users No Unstated
Clixad 00:00 UTC daily, on the earning cap rather than on the balance Yes Yes, the balance is yours

The four unpublished rows are the same four whose allowance cannot be converted into finished work. A tier that counts in sessions or in "limited" requests has nothing precise to say about the boundary either, because the quantity being bounded was never precise.

The cadence is the outage

Free tiers are sold on the size of the bucket. What shapes your week is how long you are stopped when it is empty.

The longest wait each cadence can impose, if you exhaust the allowance at the worst possible moment.
Cadence Longest possible wait Who uses it
Calendar month About 30 days GitHub Copilot Free
A day 24 hours Gemini CLI free tier, FreeBuff, Amp Free, Clixad's earning cap
Rolling five-hour window 5 hours, plus a weekly window behind it Claude Code and Codex, on paid plans
Unpublished Unknowable in advance Cursor Hobby

A monthly allowance is the most generous thing on the page and the most dangerous shape to plan around, because the failure is a three-week outage. A daily allowance is smaller and its worst case is tomorrow. Which is better depends on how you work.

The rule of thumb: match the cadence to your deadline, not to your appetite. If a piece of work has to ship this week, a tier whose reset is next month is not a free tier for that work, whatever the bucket size says. If the work is a long weekend project, a monthly bucket is exactly right and a daily cap is the thing that will annoy you.

Two things called a reset

Two different arrangements share the word, and the difference decides whether the boundary costs you anything.

A refill. You are issued an allowance, you draw it down, and at the boundary it is restored to full. Anything you did not use is gone. GitHub says this in as many words: included AI credits do not carry over between months, and unused credits are forfeited. Five of the six rows above work this way, whether or not they say so.

A cap release. You hold a balance. Something limits how fast you may add to it, and the boundary lifts that limit rather than refilling anything. Clixad works this way: a free account may earn up to 500,000 credits a day, that ceiling resets at 00:00 UTC, and the credits already in the wallet are untouched by the boundary. There is no expiry date on them.

Under a refill, a fortnight of light work funds nothing later, because the allowance was never yours to accumulate. Under a cap release, the same fortnight leaves a balance for the week the deadline lands. The cost is symmetric, though: our earning ceiling is a daily one, so an account with an empty wallet and no headroom left waits until 00:00 UTC exactly like everybody else in the table. The balance carrying over is the only difference.

What forfeiting is worth

GitHub Copilot Free grants 2,000 code completions a month. Here is what a full month of them costs at our metered rates, on the completion shape an earlier post declared in its markup: 1,200 input tokens and 40 output tokens each.

A full month of 2,000 inline completions, priced through packages/pricing and recomputed by a test. 100,000 credits is $1 of list value.
Model 2,000 completions At list price
GPT-5 Nano 24,000 credits $0.24
GPT-6 Luna 42,000 credits $0.42
Claude Haiku 5.5 42,000 credits $0.42

So a month of completions you failed to use was worth well under a dollar of inference on the models that are sensible for the job. What the boundary costs you is timing: a light month cannot fund a heavy one, so the tier has to be sized for your worst week and sits idle for most of the rest.

The 50 chat requests on the same plan are the part that bites, because chat is where agent work happens. That is about four twelve-call agent tasks a month, and it resets on the same calendar-month boundary as the completions.

The pattern runs the opposite way to what you would expect.

Anthropic's Claude Code documentation states that usage "resets on a rolling five-hour window and a weekly window", and that when a developer hits it, "The message shows when the window resets". OpenAI's Codex pricing page gives its usage estimates "per five-hour period" and adds that "Weekly limits may also apply." When this post was first written the page put it as "The usage limits for local messages and cloud chats share a five-hour window. Additional weekly limits may apply." The wording has changed since, and the window has not. Neither of those is a free tier. Both publish the shape of the window and surface the reset time at the moment it matters. OpenAI's own free plan for Codex publishes no allowance at all, as our Codex comparison found.

The free tiers, whose users can least escape a blackout by paying, are the ones leaving the clock unstated. The clearest case is Google's, where the quota page gives exact per-day and per-minute ceilings with no window attached, and a user reported in March 2026 that they had "Hit the daily quota on Friday. Error shows a reset time, but 3 days later the quota is still blocked." That issue was closed as not planned. The full reading of Google's numbers is the companion to this post.

Working through the gap

Assume the window is unknown, because for four of six it is. These moves do not depend on it.

Drop a model tier before you drop the tool. Many blocked afternoons are a frontier model doing work a small one would have finished. This only helps where cost depends on the model: on a request-counted quota every model costs one request, so a cheaper one saves nothing. On a token meter it saves almost everything. One twelve-call task is 3,051 credits on GPT-5 Nano and 230,760 on Claude Opus 5.5, and that is the same task.

Find out what the stopped task already cost. An agent that halts at step 8 of 12 has spent most of the budget and produced nothing, and that share is measurable: between 53% and 55% of a clean run, on every model in the catalogue. Knowing it tells you whether to resume or restart.

Keep a second agent installed on a different funding model. Two tiers that do not share a ceiling cannot be emptied by the same afternoon.

Prefer a balance to an allowance for anything with a deadline. An allowance is sized by somebody else for an average user, and it expires. A balance you can top up is sized by you. On Clixad that top-up is a completed advertiser action in a browser tab, credited on completion. Being screened out of a survey part way through pays nothing or a small bonus while still costing the minutes, and that is the most common outcome of starting one on any offerwall. From our own ledger: completed offers on our CPX wall have credited from 9,000 to 114,000 credits on completion, with a median of 51,000: 26 completions recorded in our ledger between 27 July 2026 and 12 September 2026, as of 14 September 2026, with payouts varying widely from one offer to the next.

FAQ

When does the GitHub Copilot free quota reset?

At 00:00:00 UTC on the first day of each calendar month. GitHub's billing documentation says included AI credits do not carry over, unused credits are forfeited, and the allowance resets to the full monthly amount. It follows the calendar month, not your signup anniversary, and it is the only free tier in this comparison that publishes its reset to the second.

Do unused AI coding requests carry over to the next month?

On GitHub Copilot, no: unused credits are forfeited. The other free tiers here publish nothing either way, which in practice means the same. An allowance refills rather than accumulates, so a quiet fortnight cannot fund a busy one. Clixad credits sit in a wallet with no expiry date, and only the daily earning cap resets.

Why can't I find out when my free tier resets?

Because most of them do not publish it. Of the six free coding tiers that include model access, only GitHub Copilot Free and Clixad state a reset window. Cursor's Hobby plan publishes no quota at all. FreeBuff and Amp Free describe a daily allowance without a window, and Google's Gemini CLI quota page gives ceilings but no reset time. These are the same four tiers whose allowance cannot be converted into finished work.

Is a daily reset better than a monthly one?

It depends on how bursty your work is. A monthly allowance is one large bucket you may spend in an afternoon, and emptying it in week one blocks you for up to three weeks. A daily allowance caps that afternoon harder, but the worst wait it can impose is until tomorrow. The reset cadence is the longest outage the tier can hand you.

What can I do while I wait for a free tier to reset?

Assume the window is unknown, because for four of the six it is. Drop to a cheaper model, which helps only where cost depends on the model: on a request-counted quota every model costs one request. Work out what the stopped task already spent, to decide whether to resume or restart. And keep a second agent on a different funding model, so one afternoon cannot empty both.

If the balance itself is the question rather than the clock, where free AI coding credits come from sorts the four sources by whether the grant returns, and the five routes to coding without paying sorts the tools by who ends up paying for the tokens.

Sources

  1. GitHub, Billing for individuals, for "Included AI credits do not carry over between months. Unused credits are forfeited, and your allowance resets to the full monthly amount at 00:00:00 UTC on the first day of each calendar month." and for Copilot Free including 2,000 code completions per month. Read 20 August 2026.
  2. Google, Gemini CLI: Quotas and pricing, for the per-day and per-minute ceilings and for the absence of any stated reset window. Read 20 August 2026.
  3. User report, Quota never resets after hitting daily limit (Gemini Code Assist for individuals), issue #22643, opened 16 March 2026 and closed as not planned, for "Hit the daily quota on Friday. Error shows a reset time, but 3 days later the quota is still blocked." A user's account of their own experience rather than a statement by Google. Read 20 August 2026.
  4. Cursor, Pricing, for the Hobby plan listing "Limited Agent requests" with no numeric quota and no stated renewal. Read 20 August 2026.
  5. Freebuff, Introducing Freebuff: the free coding agent, published 12 February 2026 and updated 22 July 2026, for "Deepseek v4 Flash only, 6 one-hour sessions per day" and for the absence of a stated reset boundary. Read 20 August 2026.
  6. Sourcegraph, Amp Free Is Ad-Free, for the 23 May 2026 note: "We're pausing or reducing further Amp Free's daily allowance for more users and restricting use of free credits by older client releases." Read 20 August 2026.
  7. Anthropic, Claude Code: manage costs effectively, for a per-seat allowance that "resets on a rolling five-hour window and a weekly window" and for "The message shows when the window resets". A paid plan, quoted here as the contrast. Read 20 August 2026.
  8. OpenAI, Codex pricing, for "The usage limits for local messages and cloud chats share a five-hour window. Additional weekly limits may apply." Also a paid plan. Read 20 August 2026, and read again on 18 September 2026, when the page gave its estimates "per five-hour period" and said "Weekly limits may also apply."
  9. Clixad's own reset: the daily earning cap is measured against the UTC calendar day in packages/db/migrations/0008_money_functions.sql, and nothing in the schema expires a credit balance. Credit figures come from packages/pricing over the model catalogue read from OpenRouter on 26 July 2026, recomputed by the test beside this post. The completion shape (1,200 input, 40 output) is read from the markup of the earlier post that declared it.
  10. Offerwall payouts: completed CPX offers as recorded in Clixad's own ledger, credited between 27 July 2026 and 12 September 2026 and read on 14 September 2026, held as CPX_COMPLETIONS in packages/pricing.