about standards, webdesign, usability and open source

HTML5 test updated: how well does your browser support HTML5 now?

Earlier today I’ve released a new version of the HTML5 test. The goal is still the same: to show an indication of how well your browser supports the upcoming HTML5 standard and related specifications.

It was clearly time for an updated test, because browsers were starting to get very close to the original maximum score of 160 points. If you disregard the codecs for a bit: a current nightly of Safari scores 95 out of 106. That is very close and demands a new challenge. The maximum of 160 was always intended to be a moving goalpost. The original test suite did not test for all of the new HTML5 features and I always intended to keep adding tests until the specification is stable and all features are properly tested.

That is exactly what happened: I’ve added a large number of new tests. If you compare the results of the original test with the new one, you may also notice some changes to the already existing tests. The original HTML5 test may award points for a feature, while the new test does not. We made some of the original tests stricter, so a browser must follow the specification more closely before points are awarded. We still don’t test all current features of HTML5, so expect more updates in the future.

The maximum score of the new test has been raised to a total of 300 points. This does not include any points for codec support. If you support one or more codecs you can get additional bonus points, but the maximum score of 300 points is in reach of every browser regardless of which codecs you support.

Safari takes the lead (for now)

The release of Apple Safari 5 yesterday marks the first time a shipping browser scores higher than 200 points on the HTML5 test. To be exact, Safari scores a total of 208 points and 7 bonus points. That is quite a milestone because Safari 4 scored 129 points and Safari 3.2 which was released just a year and a half ago scored only 29 points.

But Apple isn’t the only high performer on the HTML5 test. Given that both Safari and Google Chrome use the Webkit engine, it is no surprise that Chrome is the current runner-up.  Chrome 5 trails Safari just below the 200 on 197 points and 7 bonus points.

A bit further away are Firefox 3.6 with 139 points and Opera 10.5 with 129 points. As usual, Internet Explorer 8 concludes the list of currently shipping browsers with a meagre 27 points.

But the race hasn’t been decided yet

So far we’ve only looked at shipping browsers, but most of the browser manufacturers also release nightly builds or beta versions. These are not ready for public consumption, but should give a good idea of how well the next release of each browser will function.

We should not expect much difference from the Webkit based browser, because both Safari 5 and Chrome 5 were only recently released, but work on Mozilla Firefox 4 has been going on for quite a while. Expect a large jump from Firefox.

Unsurprisingly Safari and Chrome still hold the lead when it comes to nightly builds. Safari keeps the advantage with 220 points over the 217 points of Chrome. Both nightlies were downloaded earlier today. As expected, Firefox makes a big leap with their nightly builds to a score of 176 points and we should see further improvements before Firefox 4 ships.

And what about Internet Explorer?

Internet Explorer is an interesting case. Microsoft has been using HTML5 as a buzzword like crazy. It seems like HTML5 stands for everything cool that is happening in the browser and if you have hardware accelerated rendering, just call it hardware accelerated HTML5. If you have a nice demo for SVG, just call it a HTML5 showcase.

Microsoft even released their own HTML5 test suite in which they score 100% and all other browser score less. Of course they only test the things they actually implemented: Text selection, Foreign content and getElementsByClassName and forget about all the things they haven’t implemented.

Internet Explorer 9 Platform Preview 2 scores a meagre 32 points. Let’s hope they manage to improve on this score.

47 Responses to “HTML5 test updated: how well does your browser support HTML5 now?”

  1. Ben wrote on June 9th, 2010 at 12:18 am

    The Safari score of 208 definitely needs an asterisk. Safari 5.0 (6533.16) running in OS X may get the 208 (plus 7 bonus points), but the latest version for Windows (Safari 5.0 (7533.16)) only gets 165 points (plus zero bonus points).

  2. Andy L wrote on June 9th, 2010 at 1:26 am

    rakaz, it would be great if you could add a FileWriter API test case. Chrome 6 is expected to support it soon.

    Great work! Thank you.

  3. Matt wrote on June 9th, 2010 at 2:42 am

    Try out enabling some features in these browsers’ nightlies/betas (such as webgl support) and note the increase in scores.

  4. web wrote on June 9th, 2010 at 4:02 am
  5. [...] post: HTML5 test updated: how well does your browser support HTML5 now … Tags: a-current-nightly, because-browsers, codecs, get-very, maximum-score, original, time-for, [...]

  6. Borkis wrote on June 10th, 2010 at 12:33 pm

    Why not test the newest beta/alpha version of Opera as well ?

    Also, IE6 scores 12 points

  7. Anthony Jr. wrote on June 10th, 2010 at 2:47 pm

    Poor iPad got a 127 + 7 Bonus Points. Here’s to expecting that iOS 4 will make this number jump in the Fall. I guess we’ll see once it hits the iPhone.

  8. Jan Jaap wrote on June 10th, 2010 at 9:53 pm

    @Ben, I have Safari 5, build 7533.16, the one you mentioned installed on my Windows 7 x64 system, and it’s giving me a score of 207 with 7 bonus points.

    Maybe you have a different version of Windows in which Safari achieves different scores. But anyhow, it is different from the score you mention.

  9. Tim Sneath wrote on June 11th, 2010 at 12:23 am

    Niels, I admire your fervor in pushing for browsers to support emerging standards. As we’ve said in the past, we agree that the future of the web is HTML5; we’re deeply engaged in the standards process with the W3C; and we want to deliver a great HTML5 experience in IE9.

    That said, for real-world web developers it’s not sufficient to simply include a base-level implementation of as many standards as possible, no matter how immature or poorly adopted within the community. Even your own test has changed over the last couple of months as the momentum has moved away from technologies like Web SQL Database and Ogg Theora. If the web is to advance as much as we all want it to, it’s important that we all deliver a really high-quality, fast, consistent implementation of each part of HTML5 and associated specifications.

    The test center is our attempt to do that – to produce an exhaustive set of test cases for specifications we ourselves are implementing in IE9. These were built by our engineering team; we could have chosen to keep them in-house but instead we decided to share them with the web community at large to help others take advantage of our work and increase consistency on the web. Of course we’re going to score 100%, since these are tests we built to validate our own product – that doesn’t mean that we think our job is done.

    The fact that other browsers don’t score 100% doesn’t mean that they’re poor browsers, but it does show that there’s a little more subtlety to adopting web standards than a binary “supported or not” checkbox.

    I think a lot of people are surprised when they check out our testdrive site at http://www.ietestdrive.com. If HTML5 is going to be powerful enough to displace Flash, then all browsers will need to be able to deliver this level of performance.

    We may not be perfect, but we think IE9 is looking pretty good on those terms right now; we look forward to a spirited competitive fight with other browsers.

    Best wishes,

    Tim Sneath | Microsoft

  10. Richard Barnet wrote on June 12th, 2010 at 6:30 am

    Why won’t you support/detect Google Chrome Frame? Seems like a no-brainer…

  11. Dan Atkinson wrote on June 12th, 2010 at 11:13 am

    I’m running Chrome 6.0.427.0 and am returned a score of 217 (and 10 bonus points).

  12. Electromikey wrote on June 13th, 2010 at 6:35 am

    Got a score of 207 (plus seven bonus) using Safari 5.0 (build 7533.16) on Windows 7 Professional x64, here.

  13. Kyoish wrote on June 13th, 2010 at 8:32 am

    I’m running Chrome 6.0.427.0 dev and am returned a score of 217 (and 10 bonus points).

  14. Mark wrote on June 14th, 2010 at 4:23 pm

    Opera 10.60 Latest Snapshot:

    and 7 bonus points

  15. Wambo wrote on June 14th, 2010 at 7:21 pm

    186(+9) Points for current Win32 Firefox nightlies with “webgl.enabled_for_all_sites” set.

  16. Stratoukos wrote on June 14th, 2010 at 8:39 pm

    Opera’s latest snapshot for 10.60 jumps from 129+4 to 159+7.

  17. witek wrote on June 14th, 2010 at 9:59 pm

    Opera 10.60 (linux i386, build 6363) scroes as 159 (+7 bonus points).

    Previous build (6351) scored 129 (+5 bonus points).

    (This new builds mainly adds builtin webm, geolocation, online/offline, appcache and webworkers.)

  18. ombarg wrote on June 15th, 2010 at 4:20 am

    @Tim Sneath

    Please don’t lose time posting comments here. Just use that time to code IE9 and achieve more than 32 points! Thanks , keep participating.

  19. Winjay wrote on June 15th, 2010 at 11:12 am

    Highest score!!: 227 & 10 bonus

    Chrome 6.0.427.0 dev with –enable-webgl parameter.

  20. Jug wrote on June 16th, 2010 at 8:59 pm

    A Microsoftie recommending a Microsoft-developed test instead of an independently developed one? What a shocker. Anyway, yes, HTML5 is still in flux, it’s an unfinished standard. But one can’t neglect the importance of Web Sockets, well-supported SVG, drag & drop, Geolocation, and WebGL support. Neglecting major features like this as important is only an admission of weakness.

  21. thepheer wrote on June 16th, 2010 at 11:24 pm

    My browser scores 129 and 4 bonus points…

  22. Irish Dafonte wrote on June 18th, 2010 at 9:12 am

    Interesting…and I agree in the most part. Keep up the great work…I will undoubtedly be back before long

  23. Mitch 74 wrote on June 22nd, 2010 at 2:01 am

    Minefield latest nightly (Gecko/20100621), got 196+9 (with WebGL enabled).

  24. Pepper Spray wrote on June 23rd, 2010 at 5:38 am

    Pepper Spray says: Opera’s latest snapshot for 10.60 jumps from 129+4 to 159+7.

  25. badger wrote on June 24th, 2010 at 1:19 am

    IE9 Platform Preview supports H.264 and WebM video yet HTML5test does not detect it!

  26. Jeffrey wrote on June 24th, 2010 at 10:37 am

    Your test has a flaw.
    It checks for codec support on canPlayType(type) returning only “probably” whilst IE9 returns “mayby” which is also in accordance with the HTML5 spec

  27. Niels Leenheer wrote on June 25th, 2010 at 8:13 am

    @badger: IE9 does not support WebM. H.264 is not detected due to a bug in IE.

  28. Niels Leenheer wrote on June 25th, 2010 at 8:17 am

    @Jeffrey: After reading the spec more closely I do think it is a violation of the spec to return “maybe”. Note the use of the word must in the spec:

    “it must return “probably” if the user agent is confident that the type represents a media resource that it can render”

    “Implementors are encouraged to return “maybe” unless the type can be confidently established as being supported or not.”

    You can’t tell me IE doesn’t know whether or not it supports H.264. It does, it knows it does, so it should return “probably”.

  29. Einlander wrote on June 27th, 2010 at 5:04 pm

    @Niels Leenheer
    Unless your using windows 7 which has a built in h264 decoder, it should return probably, since vista and lower dont come with one.

  30. Niels Leenheer wrote on June 27th, 2010 at 10:06 pm

    @Einlander: IE9 preview 9 ships with a H.264 decoder for Vista. I don’t know whether or not it uses a build in decoder for Windows 7 or uses the system codec, but H.264 video works across all platforms. IE9 also gives the same reply on both Vista and 7: ‘maybe’.

  31. Evpok wrote on June 30th, 2010 at 12:17 pm

    Firefox 4.0b1 just scored 189+9.

  32. Oli wrote on June 30th, 2010 at 4:23 pm

    Hey Niels,
    thanks for the nice tool. Ideally I’d like to see a graph for each browser showing progress by release (or even by nightly build!), but that’s lots of work. Instead I’ve just put a bunch of scores in a table for easy reference:

    I’m not going to add every score, but will try to keep it updated with major releases or when there’s a big change in score (feel free to ping me on @boblet). Fwiw

    peace – oli

    PS if anyone wants to start testing browsers and graphing the results over time… ;-)

  33. Jan Jaap wrote on July 1st, 2010 at 12:01 pm

    I just installed Opera 10.60 which has added Geolocation and some other stuff to it’s HTML5 support. I get score of 159 now with 7 bonus.

  34. TTEmiel wrote on July 4th, 2010 at 12:27 pm

    Google Chrome 6.0.453.1 dev scores 220 and 10 bonus points without the webGL flag enabled.

  35. Madspyman wrote on July 11th, 2010 at 9:31 am

    I’m using the lastest Chromium 6 daily with the -enable-webgl flag (didn’t have to disable the sandbox either) and scored 227 with 10 bonus points the test. Also I’ve tested Firefox 4 beta with webgl enabled from the about:config and scored 199 and 9 bonus points.

  36. TheK wrote on July 13th, 2010 at 2:39 am

    Another new Chrome, another new record: 231+10 not with 6.0.458 on Ubuntu.

  37. Brian E wrote on July 13th, 2010 at 4:09 pm

    Firefox 4.0b1 scores 199 +9 bonus with WebGL turned on (about:config -> webgl.enabled_for_all_sites) Requires a video driver with OpenGL 2 in beta (nVidia/ATI). Final version will have software support for older OpenGL drivers, though much slower.

  38. Adam wrote on July 18th, 2010 at 10:58 am

    What about mobile browsers (iPhone, Android, BlackBerry 6.0, Symbina, Opera Mini/Mobile). Any summary of those?

  39. Simon Hollingshead wrote on July 19th, 2010 at 11:22 am

    The newest WebKit Nightlies (on safari) get 232 (and 10 bonus) for me if I enable webGL.

  40. Prestaul wrote on July 20th, 2010 at 12:15 am

    Chrome is getting credit for supporting all of the html5 input types, but it does not appear to have many of them implemented (e.g. date, number, email, tel etc.). It is treating all of these as text inputs, which is the same thing it does if it does not recognize the type. Meaning, type=”date” is exactly the same thing as type=”foobar” in Chrome 5.

  41. Niels Leenheer wrote on July 21st, 2010 at 12:26 pm

    @Prestaul At the moment there are no specific requirements in the HTML5 spec on how form fields should be rendered. So even if a browser shows a generic text field it still complies. There are differences however how these field respond to validation and such.

  42. Jon wrote on July 24th, 2010 at 8:24 pm

    SVG in HTML5 should be mandatory!!!
    All browsers should support it!!!
    Flash and Silverlight Down!!

  43. Pier Betos wrote on July 26th, 2010 at 2:50 pm

    @Tim Sneath

    My eyebrows just twitched on your comment.

    The essence of HTML 5 standard, or any web standards for that matter, is for the browser renderers to follow it. If your browser is not that compliant, you’re not that competitive enough. You see, if you render it wrong, it trickles down to the developers also developing the web applications wrong.

    That’s the gap that the end users wanted to disappear. You (browser makers) should just compete for the benchmarks, not for the rendering standards. The standards should be any browser’s skeleton, for god’s sake. :)

  44. Walt French wrote on August 3rd, 2010 at 9:24 pm

    @Adam, I can’t comment on all smartphones, but my iPhone4 w/ iOS 4.0.1 delivered 190+7. Best as I could tell, missing features vs Safari 5.0.1 only, and no features that didn’t appear in the desktop browser.

  45. SamIAmNot wrote on August 4th, 2010 at 11:31 pm

    Any plan on adding WebIDL – http://dev.w3.org/2006/webapi/WebIDL/ to the test?

  46. Dave H wrote on August 5th, 2010 at 6:08 pm

    How did the iPhone 4 score for the ‘Elements’ tests, particularly the new section elements?

  47. Ciantic wrote on August 6th, 2010 at 12:14 pm

    IE9 Platform Preview 4 gets: 96 and 3 bonus points

    There probably have been improvements, but world is going to be a sad place if they are not going to implement none of the HTML5 Form stuffs for example.

    There are lots of rendering quirks with this preview too, e.g. sometimes the texts drops out of cleartype and looks like the text from 2000.