Skip to content
Back to blog
2026-07-17·10 min·Training

Free AI Marathon Training Plan Generator: How to Get One That Adapts

By the Pace Builder Team

We make the generator this page describes. Say so up front, judge the rest accordingly.

Most "free marathon training plan" results hand you a PDF: sixteen weeks, fixed, printed for a person who is not you and a life that does not happen. It's free because it costs nothing to give away something that can't respond.

A generated plan is different in one specific way that matters: it starts from your inputs and it changes when you do. Here's what that means in practice, and where it still isn't enough.

Generate One in About Two Minutes

You need four things, and only four:

  • Your distance - marathon, half, 10K, 5K
  • Your date - the race, or roughly when you want to be ready
  • Where you're starting - honestly, not aspirationally
  • Days per week you can genuinely run - the number that survives a bad week, not your best week

Start the generator - no card, no trial timer, nothing locked behind an upgrade. Or browse 2026 races first and let a start line pick the date for you.

What a Generated Plan Actually Builds

Pace Builder creates your personalized plan in 2 minutes.

Start training for free

The structure isn't invented by a chatbot. It's the periodisation any coach would recognise, applied to your dates:

Base. Aerobic volume, mostly easy running. Unglamorous, and the part everyone wants to skip. It's the engine.

Build. Intensity arrives - tempo work, intervals - while volume keeps climbing. The long run gets long.

Peak. The hardest, most race-specific weeks. Your biggest long runs live here, and so does most of the doubt.

Taper. Volume drops so you arrive fresh rather than merely tired. Almost nobody tapers as much as they should.

Your weekly volume curve, long-run progression and session mix come out of that skeleton and your inputs - not out of a language model improvising your mileage. That distinction is worth checking in any tool you try.

How Long Do You Actually Need?

The most common real question, and the honest answer is a range, not a number.

Starting pointMarathonHalf marathon
Already running 30–50 km/week comfortably12–16 weeks8–12 weeks
Running regularly but low volume16–20 weeks12–16 weeks
Running occasionally, no long runs20–24 weeks16–20 weeks
Not running yetA year is not pessimistic6–9 months

If your date is closer than the row you're in, the plan will compress - and compressing is where people get hurt. The right move is usually a different race, not a braver plan. A generator that never tells you that is selling optimism.

Where Generated Plans Beat a PDF

It starts where you are. A PDF assumes a starting fitness. The generator asks.

It survives contact with your life. Miss a week to flu and a printed plan just sits there, week 9 accusing you. A generated plan reshuffles what remains. This is the entire argument for the category.

It answers questions. "Why is Thursday easy?" is a reasonable question that a PDF cannot field. Ours can - you type it, you get a reason.

It logs by sentence. "18k long run, 1h42, felt strong." Done. No form.

Where They Fall Short - Ours Included

It can't see you. No generator knows you're limping. Anything that hurts belongs to a physiotherapist, not an app. We are software builders, not clinicians - that's on our about page as a limit, not a formality.

It can't promise injury-free. The evidence that load management prevents injuries is much thinner than the industry implies - the famous 10% rule failed the one randomised trial that tested it. We're not going to sell you a safety feature the research doesn't support.

It believes your inputs. Say you can run five days a week and it will build for five. Fitness level is self-reported; the plan is only as honest as you were.

It has no lab. No lactate testing, no VO2 max unless your watch estimates one and you hand it over.

Free: What That Means Here

Pace Builder is free during beta. No card at signup, no countdown, no feature behind an upgrade prompt.

The reason is not charity: the product improves the more real training it sees, and we would rather spend this phase learning from runners than optimising a checkout. Paid plans will come later, and beta users keep a lifetime discount. We'd rather write that down now than spring it on you.

Worth asking of any free running tool: what pays for it, and what happens to your data. Ours: nothing yet, and your data is exportable and deletable from Settings without emailing anyone.

Pick a Real Race

A plan needs a date, and a date is easier to commit to when it has a start line attached. Some 2026 marathons worth building toward:

Or search all 453 by month, country and distance. Every race page shows the shape of the build before you sign up for anything.

Common Questions

Is the plan actually free, or free-until-something? Free during beta: no card, no timer, no locked features. Paid plans come later; beta users keep a lifetime discount.

Can I use it without a running watch? Yes. A watch gives it heart-rate data to work with, but you can log a run by typing one sentence.

What if I miss a week? It rebuilds the remaining weeks around what's left rather than pretending the week happened. If you miss enough that the goal stops being realistic, it should tell you - and so will we.

Can I change the plan after it's generated? Yes. Tell the coach in plain language - fewer days, a different long-run day, a changed race date - and the plan updates.

Is a generated plan as good as a coach's? For structure, it's the same periodisation. For everything involving your body and your head, no - and we wrote a [longer comparison](/en/blog/ai-running-coach-vs-traditional) rather than pretend otherwise.

How is this different from a PDF plan? The PDF was written before it met you and cannot change. That's the whole difference, and it's most of what goes wrong with training plans.

Start

Generate a free plan, or read how the coaching works underneath - including which model runs it - if you'd rather know what you're signing up to first.

Sources

The claims in this article rest on these. We link the study or the body itself, not somebody else summarising it.

  1. 01Buist I, et al. (2008). No effect of a graded training program on the number of running-related injuries in novice runners: a randomized controlled trial. Am J Sports Med 36(1):33–39.
  2. 02Damsted C, et al. (2018). Is there evidence for an association between changes in training load and running-related injuries? A systematic review. Int J Sports Phys Ther 13(6):931–942.

Pace Builder creates your personalized plan in 2 minutes.

Start training for free

Running tips in your inbox

Weekly workouts, running science, and coaching tips - free.

No spam. Unsubscribe anytime.