clixadHow to get free AI coding credits in 2026
Free AI coding credits come from four places: a provider trial balance, a student or startup programme, a vendor's free tier, or an ad-funded grant. The first two are one-time and carry an expiry date. The third is usually not denominated in credits at all. The fourth comes back every day.
Sorting them matters more than collecting them, because the word credit is now doing four incompatible jobs. One vendor's credit is one message. Another's is a fixed fraction of a dollar. Ours is metered against real token usage. Only the last kind can answer the question you actually have, which is what does this finish.
The four sources
Every route to a free balance in 2026 falls into one of four shapes, and the difference that decides whether it is useful is not the size of the number. It is whether the grant returns.
| Source | Returns? | Unit granted |
|---|---|---|
| Provider trial balance | No, one time | Dollars of raw API usage, with an expiry date |
| Student or startup programme | No, one time | A larger dollar balance, behind an eligibility check |
| Vendor free tier | Yes, monthly or daily | Messages, requests or sessions, rarely tokens |
| Ad-funded grant | Yes, daily | Credits metered against real token usage |
The two "no" rows are where most published advice stops, and they are the two that cannot carry a project. A trial balance is a good way to find out whether you like a tool. It is a bad way to fund the second week, because there is no second grant and the clock on the first one is already running.
The expiry clock
The expiry terms are the part nobody reads and the part that decides whether a saved balance still exists when you need it. OpenAI's service credit terms make credits non-refundable and expire them one year after purchase or issuance, and that covers promotional credits they issue as well as credits you buy. Other vendors are considerably tighter: Framer expires top-up credits at the end of the calendar month they were bought in.
The practical effect is that saving a trial balance for a serious project is the one thing you should not do with it. A credit with a deadline is only worth what you spend before the deadline, and the developer forums are full of people discovering both facts on the same day.
Worth separating two questions that get merged. Does this expire is about the balance. Does this return is about the arrangement. A grant that returns daily makes the expiry question almost irrelevant, which is why the recurring sources are the only ones worth planning around even when their numbers look smaller.
Four units, one word
The second obstacle is vocabulary. In 2026 "credit" means at least four different things, and no conversion exists between them.
| Vendor | One credit is | Converts to work? |
|---|---|---|
| Base44 | One message, 25 a month on the free plan | No |
| GitHub | $0.01 of usage; the free plan receives none | Partly, once you know the token rate |
| Linear | A dollar of a balance you top up, expiring after 12 months | Partly, same caveat |
| Clixad | 1/100,000 of a dollar of list value, metered on tokens | Yes |
Base44's row is the clearest case of the problem. Twenty-five message credits a month sounds like an allowance until you remember that a request is not a unit of work: one ordinary agent task is about a dozen model calls, because the whole conversation is resent every round. Twenty-five messages is therefore about two tasks, and no arithmetic on the number 25 was ever going to reveal that.
The same applies to a tier counted in sessions, like FreeBuff's six one-hour sessions a day, and to Cursor's Hobby plan, which no longer publishes a figure at all. The routes to coding without paying all run into this same wall from different directions.
What a recurring grant is worth
Here is our own side of it, priced rather than described. The workload is one twelve-call agent task, the same shape the tier-limits post declared in its markup and this page reads back rather than restates. One task costs 3,051 credits on GPT-5 Nano, 43,002 on Kimi K2.7 Code and 288,450 on Claude Opus 5.
| Grant | GPT-5 Nano | Kimi K2.7 Code | Claude Opus 5 |
|---|---|---|---|
| Signup bonus, 5,000 credits | 1 task | 0 tasks | 0 tasks |
| One completed action, 40,000 to 300,000 | 13 to 98 tasks | 0 to 6 tasks | 0 to 1 task |
| Daily maximum, 500,000 credits | 163 tasks | 11 tasks | 1 task |
The middle row is a range because payouts are a range. Offerwall payouts vary by more than 2x at the same duration, so no honest surface can tell you that a given number of minutes earns a given number of credits, and this one will not. What it can tell you is what has been observed: on our wall in Germany, one completed action has credited between 40,000 and 300,000 credits.
The costs belong in the same paragraph as the numbers. Credits arrive on completion, and a screenout part way through a survey pays nothing while still taking the minutes. That is the most common outcome of starting one, across the whole offerwall industry rather than in our integration specifically. Anyone weighing 500,000 a day against a competitor's monthly allowance should weigh it against the attempts that pay nothing too.
Two rows are deliberately unflattering. The signup bonus is one task on the cheapest model and none on the other two, because it exists so the first thing you meet is the tool rather than the earning step. And a full day of earning is one task on Claude Opus 5. Frontier models are not locked behind a tier here, and they are also not cheap, and both of those are true at once.
Stacking them
These sources compose better than they compete, so the practical order is the one that spends the perishable things first.
Spend the trial balance now, on a cheap model. It expires, it does not return, and its purchasing power varies by more than 400x across a model catalogue. Spending it on a frontier model is how $5 becomes six tasks instead of several hundred.
Claim a programme balance if you are eligible, and treat it the same way. Student and startup credits are larger and behave identically otherwise.
Keep a free tier for the work it suits. An allowance counted in inline completions is genuinely generous for inline completion work, and useless for agent work. That is a statement about the unit rather than about the vendor.
Use a recurring grant for anything that has to still be running next month. This is the only category where the arithmetic in the table above stays true in week four, and it is the reason Clixad meters in tokens rather than in requests. Knowing what happens when the meter stops matters as much as knowing what the balance was.
FAQ
Where can I get free AI coding credits?
From four places, and they behave very differently. Provider trial balances give you a one-time amount, typically around $5, which expires. Student and startup programmes give a larger one-time amount behind an eligibility check. Vendor free tiers give you a recurring allowance that is usually counted in messages, requests or sessions rather than in credits. Ad-funded agents give you a recurring grant denominated in credits, earned by completing an advertiser action. Only the last of those both renews and can be converted into finished work, which is why the four are worth telling apart before you plan around any of them.
Do free AI credits expire?
Usually yes, and the terms are easy to miss. OpenAI's service credit terms make credits non-refundable and expire them one year after purchase or issuance, which applies to issued promotional credits as well as purchased ones. Several vendors are stricter: Framer's top-up credits expire at the end of the calendar month they were bought in. An expiry date turns a balance into a deadline, so a trial credit you are saving for a serious project is a credit you may never spend.
Why can't I tell what my free credits are worth?
Because the word credit now covers at least four incompatible units. Base44 sells message credits, where one credit is one message. GitHub prices an AI Credit at $0.01, and its free plan receives none of them. Linear sells a dollar balance you top up, which expires twelve months after the purchase date. Clixad meters credits against real token usage at 100,000 credits per US dollar of list value. Only a unit tied to tokens can answer the question 'what does this finish', because an agent task is about a dozen model calls and each one resends the whole conversation, so its cost is measured in tokens and nothing else.
How many credits does one coding task actually need?
On our engine, one ordinary twelve-call agent task costs 3,051 credits on GPT-5 Nano, 43,002 on Kimi K2.7 Code and 288,450 on Claude Opus 5. So a 5,000-credit signup bonus is one task on the cheapest model and none on the other two. One completed advertiser action, at the low end of what our wall has been observed to pay, is 13 tasks on the cheapest model and none on most of the catalogue. The spread between the cheapest and the dearest model is larger than any allowance difference between vendors, which makes the model you pick a bigger lever than the tier you are on.
Are ad-funded credits worth the time they take to earn?
That depends on what your time is worth and on which model you run, and the honest answer includes the failures. Credits arrive on completion, and being screened out of a survey part way through pays nothing while still costing you the minutes, which is the most common outcome of starting one and is normal across the offerwall industry. Payouts also vary by more than 2x at the same duration, so no surface should tell you that a given number of minutes earns a given number of credits. What can be said is the observed range: one completed action on our wall in Germany has credited between 40,000 and 300,000 credits.
Sources
- OpenAI, Service credit terms, for credits being non-refundable and expiring one year after purchase or issuance. Summarised rather than quoted: the page refuses automated retrieval, so this is stated from OpenAI's published terms as reported by their help centre rather than from a verbatim read. Checked 13 August 2026.
- GitHub, Copilot plans and pricing, for "2,000 completions per month", "50 chat requests (including Copilot Edits)", "No credit card required.", and "1 AI credit = $0.01 USD" on the plans that receive credits. Read 13 August 2026.
- Base44, Pricing, for the free plan's "25 message credits/mo". Read 13 August 2026.
- Linear, AI credits, for "Funds you add expire 12 months after the purchase date" and a $10 minimum ad-hoc top-up. Read 13 August 2026.
- Framer, How AI credits and Agent pricing work, for "top-up credits are a one-time purchase and expire at the end of the calendar month in which they're bought" and "any credits you haven't used don't carry over". Read 13 August 2026.
- Credit figures and task costs: Clixad's own pricing engine,
packages/pricing, over the model catalogue read from OpenRouter on 26 July 2026, recomputed by the test beside this post. The twelve-call task shape is read from the markup of the earlier post that declared it. - Offerwall payouts: amounts observed on Clixad's own CPX wall in Germany, credited on completion, July 2026.