[HEAD TO HEAD]
TrainingPeaks vs Intervals.icu: Paid vs Free
Is Intervals.icu better than TrainingPeaks?
For self-coached load analysis, yes — deeper modelling, more visible arithmetic, custom fields, and free.
[THE SHORT ANSWER]
There is a free tool built by one developer that models endurance load better than a paid platform used by professional coaches. That is genuinely the situation, and the reason it is not the whole answer is that TrainingPeaks is not really selling load modelling.
The single question that decides it
Are you coached?
If yes, TrainingPeaks, and the rest of this page is background. Your coach writes there, your compliance is reviewed there, and fighting that is not worth the friction.
If no, Intervals.icu, and you should switch today because it is free and better at the job you are using TrainingPeaks for.
Why Intervals.icu outmodels the paid product
You can set the constants. CTL and ATL are exponentially weighted averages with time constants — conventionally 42 and 7 days. Those defaults were chosen for road cycling decades ago. Intervals.icu lets you change them and see what happens. TrainingPeaks presents the output as a finished number.
Custom load metrics. You can define your own, which matters enormously the moment your sport is not one TSS was designed for.
Power and pace curves that are actually explorable. Any duration, any date range, compared against any other period.
HRV and wellness tracking, integrated and free. Including sleep, resting heart rate, weight and subjective fields, in the same chart as the load.
It shows its work. The formulas are documented and the forum where the developer explains them is public. Why that is worth more than a prettier chart.
Why TrainingPeaks still costs money
The coach–athlete workflow. A coach writes structured sessions, they appear on your calendar, you execute, the file returns, and both of you see compliance and load side by side. Nothing else does this as well. This is the product.
The plan marketplace. Buy a real coach's twelve-week plan and have it populate your calendar. Intervals.icu has nothing comparable.
Polish and reliability. It is a business with support, uptime commitments and a mobile app that works. Intervals.icu is a labour of love with a functional interface and a single point of failure.
Institutional acceptance. Coaches, teams and federations use it. If you are entering that world, you will meet it.
The comparison
| TrainingPeaks | Intervals.icu | |
|---|---|---|
| Cost | Premium subscription | Free |
| CTL / ATL / TSB | Yes, fixed presentation | Yes, adjustable constants |
| Custom load metrics | No | Yes |
| Power and pace curves | Good | Better, more explorable |
| HRV and wellness | Premium, basic | Free, integrated |
| Shows its arithmetic | Partly | Yes, documented |
| Coach–athlete workflow | Best in class | None |
| Plan marketplace | Yes | No |
| Device ingestion | Broad | Broad, including from Strava |
| Mobile app | Yes | Web-first |
| Long-term risk | Corporate | One maintainer |
| Strength training in the model | No | No |
The risk worth naming
Intervals.icu depends on one person. There is no funding, no team and no succession plan. It has been reliable for years and that is not a guarantee. Export your data periodically; that is the whole mitigation and it takes a minute.
TrainingPeaks carries the opposite risk: a company that can reprice, deprecate or reposition, which it has done before.
What neither of them models
Resistance training, at all. This is worth stating precisely because both platforms appear to accept it.
TSS is defined as intensity relative to an endurance threshold — FTP for cycling, threshold pace for running. Both terms in the equation presuppose a sustained aerobic effort. A set of eight squats at RIR 2 has no threshold, no normalised power and no intensity factor.
So a hybrid athlete's lifting either does not appear at all, or appears as a TSS number they made up — which then flows into CTL, and every subsequent decision is made against a chart containing invented data. The full argument, with what to use instead.
The practical failures arrive on the weeks that matter most:
- A heavy lower-body week plus a threshold run reads as an easy week in TSB terms.
- A bike deload while lifting volume climbs reads as recovery.
- The ramp rate with the best injury evidence behind it is being computed on part of the load. How to compute it on all of it.
And the corollary on the strength side: neither platform will ever tell you your weekly hard sets per muscle group, which is the strength equivalent of CTL and the number that actually governs whether you are progressing. Count it here.
That is not a flaw in either tool. Both are endurance platforms and they are excellent ones. It is the seam every hybrid athlete ends up bridging by hand. What it looks like when both halves sit on one ledger.
Questions
Is Intervals.icu better than TrainingPeaks?
For self-coached load analysis, yes — deeper modelling, more visible arithmetic, custom fields, and free. For a coach-athlete workflow, no, and it is not close: TrainingPeaks owns plan delivery, compliance review and the coaching business layer that Intervals.icu has no version of.
Is Intervals.icu really free?
Yes, with an optional small donation that unlocks nothing essential. It is maintained by one developer, it has no venture funding and no upsell, and that is both why it is good and the main risk in depending on it long term.
Does Intervals.icu have CTL, ATL and TSB?
Yes, with more control over them than TrainingPeaks offers. You can set your own time constants, define custom load metrics, and see how the numbers are produced. TrainingPeaks presents the Performance Management Chart as a finished product; Intervals.icu presents it as a model you can adjust.
Can I use both?
Easily, and many people do. Intervals.icu pulls from Strava and Garmin, and it can read a TrainingPeaks-planned week. If you are coached, the coach works in TrainingPeaks and you analyse in Intervals.icu at no cost.
Do either of them handle strength training?
Neither, in any meaningful sense. TSS is defined as intensity relative to an endurance threshold, and a set of squats has no threshold and no normalised power. Both will let you log a strength session as a duration with a TSS figure you invent, which then pollutes the load model with a number nobody measured.
[KEEP READING]
[COMPARE]
Base vs TrainingPeaks for Hybrid Athletes
If you have a coach who lives in TrainingPeaks, that outweighs everything on this page. If you do not, the load model has a specific blind spot.
[COMPARE]
Base vs Intervals.icu for Hybrid Athletes
For endurance analysis it is genuinely excellent and free. The gap is the same gap every endurance platform has.
[COMPARE]
TrainingPeaks Alternatives, Compared Honestly
One of the alternatives is free and models load better. The question is whether load modelling is what you were missing.