Your chosen bike frame and wheelset both affect speed in Zwift, so it makes sense for racers to do a little research and choose the best tool for the job.
Frame Tests – Introduction
While Zwift’s 4-star rating system for weight and aero is simple and easy to understand, performance varies even among frames with the same star ratings–because the stars are just an approximation.
So we’ve ranked all the standard road frames (TT frames are ranked here) against each other based on their actual performance on flat roads (2 laps of Tempus Fugit) and a long, steep climb (Alpe du Zwift).
As you will see, some frames perform well in flat tests, while others are nimble climbers. In between we have the all-arounders: frames that perform well on all course types but can’t beat the specialized frames on their own turf.
See the complete list of frames available on Zwift, along with level requirement, star rating, and Drops price >
Charts last updated April 19, 2022
This chart shows actual timings for flat and climb tests, in seconds. The shorter the bar, the faster the frame.
- Flat tests run on 2 laps of Tempus Fugit at 300 watts steady with an isolated 75kg rider 183cm tall, using basic Zwift 32mm Carbon wheels. This test data is used for all flat results on this page.
- Climb tests run on Alpe du Zwift using 300 watts steady with an isolated 75kg rider 183cm tall, using basic Zwift 32mm Carbon wheels. This test data is used for all climbing results on this page.
Frame Percentile Rank, Stacked
This format lets you easily see how frames perform across both flats and climbs, and compare that performance with other frames. This is useful for selecting the best bike frame for your race course.
Example: you want a top-performing frame for a road race that ends on the Innsbruck KOM, so you look at the longest bars. But you know the race will be won or lost on the big climb, so you go with the Scott Addict RC since it is near the top of the climbing rankings and also offers decent performance on the flats.
Frame Time Improvement, Stacked
This format lets you easily see how frames perform across both flats and climbs in terms of actual timing, and compare that performance with other frames. This is useful for selecting the best bike frame for your race course, and may be an improvement over the percentile chart above since it is based on precise timings and not percentiles.
To be precise, this chart shows how many seconds a particular frame improves on the time of the lowest-ranked frame for both the flat and climb tests. So the Scott Addict RC’s flat improvement of 23.5 means it is 23.5 seconds faster than the Zwift Buffalo on our flat test.
Since there is a bigger swing in climb times (130.5 seconds) than flat times (29 seconds) it could be argued that this chart is overly weighted toward climb performance.
What About the Tron Bike?
The Tron bike (Zwift Concept 1) is not listed above because it’s impossible to test the Tron frame without also testing the Tron wheels, since they are inseparable.
What we do know is the Tron is not a strong climber, but it’s among the fastest setups for flat and rolling courses. Only a handful of frames, coupled with the disc wheelsets, can beat the Tron by 1-3 seconds in our ~50 minute flat test.
Here’s a post comparing the Tron bike against top-performing frames and wheelsets >
We’d love to know what you think of these charts, including suggestions for improvements. Share your comments below.
That’s a different list! https://zwiftinsider.com/charts-tt/
It’s here–we just call it the “Specialized Tarmac” here, since that’s what Zwift calls it. There’s also a “Tarmac Pro” which you unlock at Level 11 — that’s the one you want. Best climber in game!
Your charts have the Spec Aethos as the best climber in the game. And the Cann EVO out-climbs the Spec Tarmac Pro at a lower level and with fewer drops. What am I missing?
I downloaded the excel sheet (Frame Timings) and just wondering what format those times are in?
Ie. the S-Works Venge has a flat time of “3078” and a climb time of “2948”
Are those numbers just total seconds?
Yep, total seconds.
It’s a TT frame – those have their own set of charts: https://zwiftinsider.com/charts-tt/
Go with the most aero setup – 808 over 3.4. Allez over Tarmac.
Hello, great analysis, thanks for the info. I was wondering how a climbing bike with aero wheels compares to an aero bike with climbing wheels. This is for rolling terrain. Which has the larger impact, frame or wheels? Thanks for all the hard work you put in!
Wheels have the larger impact… around 3-4x more impact than the frame.
I really lucked out. Won the Milinsteins on my very first time up the AdZ! Always put them on for any route with lumps in it.
It’s nothing really special… around the Zwift Aero on flat roads. Not a top climber.
Thank you for the reply. So not much different than the Parlee ESX. Thank you for taking the time to do this. This information is super useful. Too bad there is not data like for real life cycling :).
Chad, I think everyone would.
I have a feeling Eric is sworn to secrecy on the Tron bike though 😛
You must be reading the charts wrong. The Tarmac Pro finished the Alpe in 2933 seconds. The fastest TT frame (Spec Shiv S-Works) does it in 3003.
Thank-you. I thought I must have something wrong. Working on a question and can’t conclude. Thoughts welcome – I have just under 540k drops and I’ve just turned L12. That leaves me two or three rides short of a Cannondale Evo plus a Zwift Aero OR do I just go for the Spec. Tarmac Pro. I currently have a Zwift Carbon and, from prizes (I think it was) Zipp 808 and 202 wheels so nothing to save for there. I think it comes down to ‘does the Zwift Aero give a worthwhile sprinting advantage?’.
True. But also worth noting: increasing your power by, say, 2% doesn’t mean your time decreases by 2%. You get diminishing returns the faster you go, which is why, at the high end of bike performance (both on Zwift and outdoors) a 2% speed increase can be a big deal!
Just added it (and the other 3 recently-added frames today). You can see it above!
The text above the charts explains the distance/time. X axis is either seconds or percentage, depending on the chart.
Fixed now. 88% is the correct figure at this time for flat percentile rank.
Yes, it’s there now!
Yes, Spec. Tarmac is the one you get for completing the California Challenge. Spec Tarmac Pro is a much better bike.
It’s there now.
It’s already in the list above. Also, I wrote about it (and three more newish frames) here: https://zwiftinsider.com/new-frame-performance/
Don’t know if you’ve got the data that you needed but I’ve re-compiled the data in Google Sheets with Level and Drops needed. I didn’t add stars though because I think it’s less useful than actual test data that Eric has collected. https://docs.google.com/spreadsheets/d/1baYiRNVstt-VYNSBazf2hxwf2MUXuT08M1m9eq3G4Hg/edit?usp=sharing
Great work! I like the way you calculated the combined rating. I think that’s fair: 50% of crunch time on the flats, 50% of crunch time on climbs. I’m very interested in how the new Specialized SL7 scores. Cannondale Evo and Trek Emonda are under-rated, generally.
Well, I guess that’s the intention of the author too since the course selection is takes roughly the same time to complete. If you look at the distance traveled though, it becomes clear that the combined ranking is only usable when a course is around 25% steep climb (around 8.5% like alpe) and 75% flat. Personally I would aim to get the best-ranked climber and sprinter frames and wheels and switch them depending on the course.
Venge S-Works (fastest on the flats) has the same issue
It’s the weird way Excel does percentile rankings.
Working on speed tests now! It was only released about 2 hours ago… 🙂
oh! haha! Sorry Eric, i didn’t realise…got the email through advertising a ride with that bike, and looked. Didn’t realise it was so new! Apologies 🙂
And I’m level 28.
Nah, bike frame choice doesn’t affect your lack positioning in any way.
About a year ago, Eric said the Tarmac Pro is the one you want. It is further up the page.
Ah, good catch! Just updated this yesterday, but used the wrong times for those (those times were with a faster wheelset). Fixed!
Still working on the climb test data for frames. It takes a while, since we have to do two Alpe climbs for each frame, and there are around 90 frames! Got around 25 frames left…
Flat and climb tests are done for wheels, and flat tests done for frames.
Awesome! Much appreciated!
Similar to the Pinarellos … see my comment below. This way Zwift are not making friends with the manufacturers.
The Dogma F has the same overweight groupset issue that other new bikes have… see https://zwiftinsider.com/groupset-upgrade-cervelo/
Nothing new for TT frames and wheels that I’ve discovered.
Thanks for that information!
For the second issue, did you update the climb timings? For the Cervelo S5 2020 it is still 2948, which should be the pre-groupset-change number, if i am not mistaken…
I am refering to the csv file..
Just fixed the S5 2020’s climb time.
Not since an update last autumn
Yeah, I worked for and got that frame two months ago thinking the same thing, as I had researched the crap out of how to best spend my drops. And just a couple days after that, I read for the first time about the Dura-Ace crankset nerf (since who knows if it’s a bug Zwift intends to fix?). To see how severely it affects its climbing performance is disappointing. As I’m only lvl. 12, I’m not sure what the next best all-rounder would be… Good thing I’m 25% along toward the Tron bike, as all this is essentially moot once… Read more »
The raw data provided here made for such a treat. Good fun. One thing that strikes me is how wide the spread of values is for climbing, as opposed to flat, times. Without the 3 Zwift outlier frames, Tempus Fugit times have a median around 3092.5 seconds. The very best sprint bike trimmed 14 seconds off that measure of center, in two full laps. The very worst bikes — again, no Buffalo-levels included — are not more than 5 seconds above the median time. There’s a “middle class” over half the distribution, but no real poverty when it comes to… Read more »
The S5 2020 got nerfed by the Dura Ace “weight” bug where it loses a ton of time on the climb (Along with some other favorites, specifically the Canyon Aeroad ’21 and Dogma F). https://zwiftinsider.com/groupset-upgrade-cervelo/
Some rumors and winks that this has been fixed in the recent update, anxiously awaiting updated tests to see.
meant to say “drive-train weight bug”… not DA specifically, given the S5’s on eTap, lol
So then. Top 5 Aero frames: 1. Venge S-Works (lvl 33) 2. Felt AR (lvl 34) 3. (tie) Uranium (lvl 19) 3. (tie) Cervelo S5 2020 (lvl 27) 5. Aeroad 2021 (lvl 10) Top 5 Climbing frames: 1. Aethos (lvl 34) 2. Chapter2 TOA (lvl 18) 3. Can. Evo (lvl 9) 4. Addict RC (lvl 33) 5. Cuba C:68X (lvl 32) Top 5 Allrounders: 1. Chapter2 TOA (lvl 18) 2. Addict RC (lvl 33) 3. Dogma F (lvl 32) 4. Aeroad 2021 (lvl 10) 5. Venge S-Works (lvl 33) It worth noting that the Aeroad 2021 is the best Allrounder… Read more »
If I got Erics recent post right, the Chapter2 has already lost its crown of the allrounder and the Aeroad is back on the top. Eric, can you shortly update the new TOA numbers, please? 🙂
Do you know if these charts are up to date as of 3/21/2022 (when Erik updated the post that brought up this issue, to say that it was a testing error and not an actual update to the frames)? In other words, is the “old” data from earlier this month still accurate?
Yes, these are up to date.
So does that mean that the TOA is still top dog in spite of previous reports that it had been knocked down a peg or two?
If you push lower watts, the time gaps between frames will grow. Higher watts, they shrink.
Read how height affects speed here: https://zwiftinsider.com/speed-tests-height/
Lights vs aero depends on a multitude of factors, especially which course you’re riding.
I don’t think they are. This chart is updated, however, as far as I know:
Hi Eric. Lovely stuff, as always. I can’t help but think, however, that these comparisons miss the point slightly. It’s interesting how well frames and wheels perform on their own, but the key to Zwift racing is how they perform in the draft of a pack of riders. i feel the most useful data would be “how many extra watts, (compared to a baseline of, say, the steel bike) does each bike need to stay in a pack of, say, 15 riders.” For a lot of bikes, I would guess that actually the difference would only be 1-5 watts. (Which… Read more »
You are correct. If the flat times without drafting roughtly correspond to the performance when you are drafting on a percentage basis then comparing the most aero frame (Venge at 3077 flat time) to the best climber (Aethos at 3093 flat time), at 250 w you’d require 3093/3077*250 = 251.3 w to go the same speed as the Venge if you were riding the Aethos.
Eric’s frame timings spreadsheet has the Spec Aethos climb time at 2930 and the Spec Tarmac Pro at 2936. Wouldn’t that make the Spec Aethos a better climber than the Tarmac Pro?
The Aethos is probably a better climber irl so that makes sense. No?
This page takes FOREVER to load. Maybe because there are heaps of comments, oh dang, I’m just making it worse
You have to force stop load on the page. Something is hanging up the page. Once you do that, the charts show up. At least it does for me
Charts aren’t showing?
Hi , first of all thank you for the stats and the time you put in all this . I have a little request , I like the Scott foil 23 but I don’t see it in your stats list . I don’t know if Zwift added that bike but will by cool to see we’re he end on that list .
Hi, I believe the charts have some hickups. While the trek madone shows 2398 seconds climb (faster than Canon Aeroad 2021) in the excel table, in the bar chart it’s 2952 seconds (slower than Canon Aeroad 2021)…
also, it differs a bit from the table found here: https://docs.google.com/spreadsheets/d/1gvSpoGFk4NSuYKBLUKlUwl9VJomd1zYH9nUE-ItKM5c/edit#gid=1825340343
I’m not seeing those differences… I see the Madone as showing 2952 everywhere.
Hi Eric! Not that it is a huge difference, but the frame performance scatter plot puts the TOA at 2934 secs climb, while the frame timings list in this page has it at 2943. One of the numbers seem to be a typo, could you clarify that?
What happened to the Chapter2 KoKo? Where does it fit?
It hasn’t been added to these charts yet, mostly because its performance isn’t anything to write home about.
You can read about its performance here: https://zwiftinsider.com/chapter2-koko/
Really sorry to be an idiot, I can’t get the CSV or Excel download link to work? I’d love to be able to rank the frames in Excel. Are the links working ok or is it me?
My reading is s5 2020 is fastest flat performer?? I thought it was venge. Am researching for wtrl round 4 sprints stage on new flats. Am level 41 so was thinking venge and 858s but this looks like S52020….confused now.
There are four frames that are all basically the same performance: Cervelo S5 2020, Specialized Venge S-Works, Uranium Nuclear, and Felt AR.
You can see that in the first chart on this page.
The Venge and AR actually climb just slightly faster than the other two, but I consider them all to be basically equal. Pick the one you like the most!