Later AI Pricing: How Far the Caption and Idea Credits Go on Each Plan

Later costs $18.75, $37.50 or $82.50 a month on its Starter, Growth and Scale plans when billed yearly, with 5, 50 and 100 AI credits a month. One credit writes one caption or suggests three ideas, so Starter covers about five captions a month, Growth about fifty and Scale about a hundred.

The credits are the smallest part of what you pay for, and the easiest to run out of. Starter's five credits would not cover a single week of daily posting, which is why the useful question is not "how much AI do I get" but "how many of my posts need AI at all". On Growth and Scale you can buy more; on Starter, five is the ceiling.

Follow me on Instagram@sagnikteaches

The three plans side by side

The figures below come from Later's pricing page, read on 5 November 2026. Prices are USD per month on yearly billing; the page also has a monthly-billing toggle, which I come back to further down.

Connect on LinkedInSagnik Bhattacharya

Two things follow at once. The AI allowance grows far faster than the price: Growth costs twice Starter and gives ten times the credits. And the 30-posts-per-profile cap on Starter matters as much as the credits, because a business posting daily on Instagram uses 30 a month on that one profile. A social set, in Later's terms, is one profile on each platform it supports (Instagram, Facebook, TikTok, Threads, YouTube, Pinterest, LinkedIn and Snapchat), so a business with two Instagram accounts needs two sets, not one.

Subscribe on YouTube@sagnikteaches

What a credit buys

The page says a credit generates one caption or three ideas. The two features are the Caption Writer, which drafts an on-brand caption for an Instagram post, and Ideas, which produces content pillars and post ideas. Each press of the generate button uses a credit. That wording has a consequence people miss: asking for a different version of a caption is another credit, not a free tweak.

So the unit worth counting is the generation, not the post. A post you caption once costs one credit; a post you reroll three times costs three. A batch of ideas costs a credit and returns three of them, which makes ideas the cheap end of the allowance. Both features draw on one pool, and Later's help centre says the pool resets on the 1st of every month at midnight UTC, whatever date your plan renews. When it is empty, the AI features stop until the reset or an add-on.

Worked example: a garden centre posting 20 times a month

Say a garden centre posts to Instagram and Facebook about five times a week, which is roughly 20 posts a month, using the same photos on both. It has one Instagram profile and one Facebook page. The owner wants to know which plan to pick and whether the AI credits matter.

First the post counts. Twenty posts on each of two profiles is 20 scheduled posts per profile, inside Starter's 30 per profile. The posting cap is not the problem; the credits are.

Now the credits. Suppose the centre wants Later to draft every caption. The Caption Writer is described for Instagram posts, so the owner generates one Instagram caption per post and adapts it for Facebook by hand. With one attempt per caption that is 20 credits. With the realistic habit of rerolling once for about half the posts, it is 30. Add one idea session at the start of the month for seasonal themes, which is one credit for three ideas; say four sessions across the month to cover bedding plants, tools, a workshop and gift vouchers, for four credits. The month comes to somewhere between 24 and 34 credits.

If every caption must come from Later, Growth is the plan, and it costs $18.75 more than Starter. Dividing that gap by the 45 extra credits gives about 42 cents a credit, but that is a poor measure, because Growth also adds a second user, a second social set and a longer analytics history. A fairer reading is that Growth's AI allowance is the headline reason to move up only if captioning is where your time goes.

The cheaper route the owner tries

The centre's owner points out that the five Starter credits are enough for ideas if she is careful: five ideas sessions give fifteen post ideas, which is most of a month's topics. She writes the captions herself, using a general chatbot she already has, and uses Later only to schedule. On that plan the AI cost inside Later is zero beyond the first five credits, and the bill stays at $18.75. The trade is that her captions come from a different tool and do not use whatever brand voice Later's writer has learned.

That is a legitimate choice, and it is the one I would test first. If the chatbot route turns out to take twenty minutes a week more than she wants to spend, she can move to Growth for the month it matters.

Counting your own month in four steps

1. Count posts per profile. Add up what you schedule on each profile in a typical month. If any single profile exceeds 30, Starter is out regardless of AI.
2. Decide where AI helps. Captions for product posts, yes; replies to customers or anything personal, no. Mark only the posts where a draft would save time.
3. Estimate generations, not posts. Multiply captioned posts by your honest attempts per caption (1.5 is a fair guess), then add one credit per ideas session.
4. Add a third of headroom. Seasonal months run heavier. If the answer is 24, plan for 32.

Compare the result with 5, 50 and 100. Almost every small business lands in one of two places: under five, where Starter is fine and AI is a nice extra, or between 20 and 45, where Growth fits. Scale's 100 credits are rarely the reason to pick Scale; its 48 profiles, four users and unlimited posts are. If you need more than 50 credits but none of Scale's extras, Growth plus one 100-credit add-on gives 150 credits for $41.25 a month at yearly rates: more credits than Scale for about half its price.

A filled-in month: where the 32 credits went

To make the credit count concrete, here is a month for the garden centre, written the way I would keep the log. It is illustrative, but the pattern is realistic.

That adds up to 28 credits, inside the 24 to 34 range estimated earlier. The telling detail is the rerolls: seven of the 28 credits went on regenerating, a quarter of the allowance. If the owner had written one fuller request per caption, the same month would have been about 21. On Growth that difference is invisible. On Starter it is the difference between running out and not, and it is the reason to learn to ask well.

A request that gets a postable caption first time

Whatever direction Later lets you add to a caption request, the structure below keeps credits from vanishing on vague drafts. If you draft in a general chatbot instead, the same structure works there.

Write one Instagram caption for a garden centre post.
Photo: three trays of lavender, price tag visible.
Offer: 3 trays for the price of 2, Saturday only, until 4pm.
Tone: friendly, plain, no exclamation marks.
Length: two sentences plus a line asking people to tag a friend with a garden.
Hashtags: three, all about lavender or bedding plants.

A realistic first result (illustrative): "Lavender season is here, and this Saturday three trays cost the price of two. Offer ends at 4pm, so come early for the best pick. Know someone with a sunny border? Tag them below. #lavender #beddingplants #gardeningtips". What the owner would fix: "come early for the best pick" is a claim about stock levels the centre may not want to make, and the third hashtag is generic. Two small edits, no reroll, one credit spent.

Compare a vaguer request that cost three credits in the log above: "Write a caption about our sale". It returned a breathless line with three exclamation marks and a discount percentage the centre was not offering. The credits were spent finding that out. Specific input is the cheapest way to save them.

Ideas credits are the better bargain

Because one credit returns three ideas, the idea generator gives more output per credit than the caption writer. The risk is the opposite one: ideas come back generic, so a credit spent on "ideas for a garden centre" returns "show your plants". Give it a constraint instead and the three ideas become usable. A filled-in request for a pet shop would be: "Three post ideas for a pet shop in October, using photos we already have of dog coats, a rabbit hutch and a cat tree, each with a reason to post on a weekday". The output is still a starting point, but each idea is tied to stock the shop owns.

A practical rhythm for Starter's five credits is to run ideas once a week for the first three weeks (nine ideas, three credits) and keep two credits for captions on posts you are stuck on. For a longer take on planning a month of content, planning a month of barbershop Instagram content with AI shows the same method on a different business.

A mistake that turned five credits into one week

Here is a pattern worth recognising, as an illustration. A pet shop owner on Starter used the caption writer for a launch post, did not like the first result, and pressed generate four more times chasing a better one. Five credits gone by Tuesday of week one. The captions were not bad; the instruction was. She had asked for "something fun" about a new range of rabbit toys, and each draft was fun in a different direction. She finished the month writing captions by hand and decided the AI was not worth it.

The fix is not a bigger plan, it is a rule: no more than two generations per post, and if the second is wrong, change the instruction, not the dice. Written down on a card by the desk, a rule like that stops one hard post eating the month. If a plan upgrade would only be used to fund rerolls, the upgrade is paying for the habit rather than the work.

Four quick cases for other shops

A pet shop posting three times a week

Twelve posts a month on Instagram only. Captioned with one attempt each, that is 12 credits, which is more than Starter's 5 but nowhere near Growth's 50. If the owner writes half the captions herself and uses the writer on six, she needs 6 to 9 credits. Starter with a bit of discipline, or one month on Growth around a sale.

An optician with a quiet feed

Eight posts a month: new frames, a care tip, a staff photo, an eye-test reminder. Here AI adds little, because the posts are short and personal. Starter covers all of it, and the five credits go on ideas for seasonal topics. Paying for Growth would be spending $18.75 a month on credits that never get used.

A veterinary practice with two staff posting

Two people share the work, one on social and one on appointment reminders, so the second user on Growth matters more than the credits. A vet practice would also be cautious about drafts for health advice, which should be written and checked by a clinician, not generated. The realistic credit use is low, but the two-user limit decides the plan.

A garden centre with three sites

Three branches, each with its own Instagram and Facebook profile, need three social sets, because a set holds only one profile per platform. That rules out Starter's single set at once. Growth has two sets; on yearly billing a third costs $11.25 a month and a third manager's login $3.75, so Growth comes to $52.50 a month with three sets, three users and 50 credits. Scale, at $82.50, has six sets, four users and 100 credits. Unless the branches want Scale's custom analytics or competitor tracking, Growth with add-ons is the cheaper fit, and at 20 posts a month per branch its 180-per-profile cap is nowhere near.

How to stretch five credits without spoiling the posts

If you stay on Starter, the allowance is best spent where a machine does something a person finds hard, and saved where a person does it in a minute. Ideas are the best value: one credit returns three, so five credits give fifteen. Captions are worse value, one for one. A reasonable split for the month is three credits on ideas (nine topics) and two on captions for posts where you are stuck.

To avoid rerolling, put more into the first generation. A caption request that already contains the product, the one fact to include and the tone usually needs no second go. A before-and-after for a plant sale post shows the idea. Weak input: "caption for a plant sale". Stronger input: "Caption for a Saturday bedding plant sale: 3 trays for the price of 2, ends 4pm, friendly, two sentences, no hashtags". The second version usually returns something postable on the first credit, and the owner's edit is a word or two.

Check every AI caption before it goes live

Whatever credits you use, read the caption against the post. Look for invented details (a discount you are not running, a feature the product lacks), the wrong day or time, hashtags that do not match the product and phrasing that is not how you speak. If you want drafts to sound like you, the technique in writing Instagram captions with AI that sound like you works in any chatbot and costs no Later credits. Whether AI belongs in your posts at all is covered in should a small business use AI to write its social posts.

Monthly versus yearly billing

The prices at the top are billed yearly, and Later labels that as 25 per cent off with three months free. The monthly price is on the page behind a "Bill Monthly" toggle. Later's own FAQ puts it as saving 25 per cent against a monthly subscription, and working back from that suggests roughly $25, $50 and $110 a month, but I have not read those numbers off the page, so check the toggle. The add-ons follow the same pattern: the help centre prices 100 extra credits at $5 a month, against the $3.75 the yearly view shows. Every plan also starts with a 14-day free trial, which is long enough to count a fortnight of real generations. After that the practical advice is simple: pay monthly for the first two months, while you find out whether you use the credits, and only then switch to yearly.

The same reasoning applies when you compare Later with other schedulers. Their entry price may look lower, but if the cheaper plan has no AI or a lower post cap, the sum is different. The roundup of AI social media tools lines them up, and what it costs to automate social media with AI adds the time side of the bill.

What to do when credits run out mid-month

If you hit zero on day nineteen, you have three options, and they are not equal. Write the remaining captions by hand, which is the right answer for most posts. Use a general-purpose chatbot for the drafts and paste them into Later, which keeps the scheduling benefits. Or buy more: on Growth or Scale a 100-credit add-on is prorated the first time you buy it and can be removed again, though removed credits stay until the end of the billing period. An upgrade makes sense only if you have a heavy fortnight ahead, such as a launch or a seasonal rush, and Later counts the credits you have already used this month against the bigger plan's allowance.

Whichever you choose, write down how many credits the month used and what you would have needed. After two months you have a real figure, and the plan decision stops being a guess. If you want help mapping which of your posts deserve AI drafting and which should stay yours, that is the sort of job my AI implementation consultation covers.

A sticky-note summary of the sums

• Credits needed = captioned posts times attempts per caption, plus ideas sessions.
• Starter (5) fits under five credits; Growth (50) fits 20 to 45; Scale (100) fits only when you also need its extra users or profiles.
• The posting cap, 30 per profile on Starter, can rule it out before AI does.
• Growth and Scale can add 100 credits a month for $3.75 at yearly rates; Starter cannot add any.
• Credits reset on the 1st of the month, so unused ones are gone.
• Pay monthly first, yearly once the numbers are real.

Later credit and plan questions

Does Later charge a credit every time I regenerate a caption?

Per Later's pricing page, one credit is used each time you generate captions, and another each time you generate ideas. So a regenerate costs another credit. Treat every press of the generate button as a spend, and edit a decent first draft by hand instead of rerolling.

Do unused Later AI credits carry over?

No. Later's help centre says credits reset on the 1st of every month at midnight UTC, whatever your renewal date. Extra credits are sold as a recurring monthly add-on rather than a one-off pack, so there is no stockpile to build. Spend the allowance on the posts that most need a draft.

What does monthly billing cost compared with yearly?

The page shows yearly prices with a 25 per cent discount and three months free, and puts the monthly amounts behind a toggle. Later's FAQ says yearly billing saves 25 per cent, which implies about $25, $50 and $110 a month on monthly billing, but that is my arithmetic. Flip the toggle and read the real figures.

Can I buy more AI credits on Later?

On Growth and Scale, yes: the pricing page lists 100 extra credits a month for $3.75 at yearly rates, and the help centre gives $5 a month. It is a recurring add-on you can remove, prorated the first time you buy it. Starter cannot add credits, users or social sets, and one-off credit packs are not sold.

Can I write captions elsewhere and just use Later to schedule?

Yes, and many owners do. Later's scheduling, media library and analytics work without touching the credits. The AI credits only matter if you want Later's own caption writer or idea generator.

Further reads

• How Florists Can Use AI for Orders, Reviews and Social Posts — Social posts alongside orders and reviews.
• AI Prompts for Bakers: Pricing, Captions and Customer Replies — Caption prompts you can reuse in any tool.
• Coffee Shop Social Media With AI: Captions, Reels and Rhythm — A month of content from one photo session.
• How to Write Facebook and Instagram Ad Copy With AI — Turn a good post into paid ad copy.
• How to Turn Instagram Comments Into Leads With Keyword Auto-DMs — Turn post comments into enquiries.
• How to Create On-Brand Social Media Graphics With AI — Graphics to pair with scheduled posts.
• Buffer vs Later for a Cafe or Bakery's Instagram: Which Plan Fits? — A cafe's Instagram week priced on Buffer and Later: post caps, AI caption limits and what each plan costs over a year.
• AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.

Sources: later.com pricing page (plans, social sets, post limits, users, AI credit allowances and add-ons) and Later help centre article on plan add-on credits (monthly reset, add-on pricing), read on 5 November 2026.