Mojo 1.0
modular.comI feel like this language would really benefit from some sort of 1-pager overview.
I just spent a fair bit of time on the official site, and I still don't think I have a very good grasp of what problem this language aims to solve, or why I would select it over other similar languages
Their communication definitely needs work. I listened to the interview with Chris on the Lex Friedman podcast a few years ago. Back when the pitch still was "minimal python superset that allows us to make it go voom". It wasn't a bad pitch. But it's gotten a lot murkier since then apparently. And obviously pitching a new language just before world plus dog was about to switch to delegating programming work to AI agents wasn't great timing.
> "Qualcomm to Acquire Modular"
That's linked from their news section. If I read between the lines here what happened is the VC money ran out and they arranged some kind of acquihire. That obviously raises a lot of questions about what will happen to Mojo and the intentions around it. The merger was completed a few weeks ago so apparently this is a done deal. Them wrapping a ribbon around the 1.0 a few weeks later is a bit suggestive.
The modular front page reads like it was trying to pitch the whole thing as a competitor to other inference platforms like open router, replicate, bedrock, etc.
I'm sure there's more to it and that the tech was amazing. But the acquisition raises a lot of questions about the future of the whole thing. An OSS move at least would allow people to continue working on Mojo independently from Qualcomm. I assume some people in the team might be moving on post acquisition.
The Python tie in was always a sales pitch. The technical pitch was "a language built around MLIR taking all the lessons learned by Rust and Swift on board, with an eye towards Python interop". There were MLIR intrinsics and extensive discussions of the ownership model from the start, but never even an attempt to have classes.
And the main use of the language (and justification for the VC money) was always the engine built with it, which I have absolutely no opinion on.
Honestly, the promise of the language is immense. Ownership semantics that have much better ergonomics and are much more teachable than Rusts, and eventually a relatively smooth teaching path from Python to Mojo would make Mojo a default first modern low level language for many people coming in. Whether this will be relevant in the LLM world we live in now, where you can just take your Python prototype and have it be rewritten in Rust, who knows...
I am already using agentic orchestration tools where the language is irrelevant, e.g. Opal, Workato, Boomi.
The evolution of iPaaS low coding tools.
Just like you can take some high level code in a dynamic language and expect the JIT compiler to do the best every time, under completly different execution patterns, so will LLM based development eventually become like that.
COBOL is already laughing at the efforts to write so much English text to keep AI on track.
> Back when the pitch still was "minimal python superset that allows us to make it go voom". It wasn't a bad pitch
Sorry, but what's the pitch?
I felt the same way, even after browsing around on the website, and somehow I got to https://mojolang.org/docs/manual/quickstart/, which brings you to mojolang.org instead, which is the real website of the language it seems, instead of "Modular" which I guess is the business.
https://mojolang.org is a lot better as a introduction to the language for a programmer, at least for me, compared to whatever I could find on the modular.com website.
Having written a lot of Mojo over the last two year, just for fun, it's a really cool language. Ownership model adjacent to Rust, comptime in the realm of Zig, rich type system, first class SIMD support, etc. Performance wise it's the first language in long time that isn't just an LLVM wrapper. LLVM is still involved, but they are using it differently than say, Rust or Zig.
Very excited for Mojo once it's open sourced later this year.
> LLVM is still involved, but they are using it differently than say, Rust or Zig.
Chris Lattner said in a Developer Voices podcast interview:
"Swift in a way was syntactic sugar for LLVM, at the very bottom of the stack it could talk directly to LLVM primitives. Mojo does basically that same trick, but it supercharges it by moving to this MLIR world."
> Performance wise it's the first language in long time that isn't just an LLVM wrapper
Why? Because it uses MLIR? Rust has its own MIR, it's even more not-LLVM-wrapper.
Probably because the creator created both LLVM and MLIR.
I'd assume Mojo does more complex transformations at the MLIR level than Rust does at its MIR level.
First, I'm not sure. Rust has MIR optimizations and also other transformations.
But even if yes, MLIR is part of the LLVM project, so it's technically still "just an LLVM wrapper".
MLIR is more a library for building a compiler than a compiler itself IIUC. LLVM can be thought of similarly, but IMO, it's more self contained.
How are compile times compared to Rust? Zig? Go?
Mojo is much faster than Rust, but slower than Go. I don't have much experience with Zig.
Mojo is designed to be very fast to compile, so many decisions were made to keep the core language simple, one example is in `where` clause, the checking there happens entirely in the parser. It can be stupid in situations where other languages are smart, and make you explicitly do checks that could be inferred in other languages. But on the other hand, it can be very fast since it doesn't have to do all the complicated resolving.
Another reason is how LLVM is used, Modular found ways to parralize LLVM code generation by introducing novel techniques that'll likely become a lot more popular with other languages.
> Modular found ways to parralize LLVM code generation by introducing novel techniques
What techniques?
I think they had a good dive here https://www.youtube.com/watch?v=SEwTjZvy8vw The gist is that MLIR (unlike LLVM) was designed to be multithreaded so the whole stack above LLVM IR is parallel, and then LLVM is used as a per-function codegen tool in parallel, so although LLVM itself is single threaded there are multiple seperate contexts of it. also I think they wrote their own linker.
That is hardly novel, here is the thesis for the Oberon multithreaded compiler from 2003.
https://www.research-collection.ethz.ch/entities/publication...
C# compiler is also fully multithreaded since the Roslyn rewrite in 2016.
The novelty is in parrelizing LLVM, which is traditionally a single-thread code generator.
So basically LLVM catching up with times.
Well, LLVM itself is still single-threaded; they just designed Mojo's compiler in such a way that the code generation part can be parallelised, which is something you can do if you understand LLVM from top to bottom. As far as I can see, no other production-grade LLVM consumer has been able to make it parallel; the single-threaded nature remains a major performance bottleneck.
> Modular found ways to parralize LLVM code generation by introducing novel techniques
> The novelty is in parrelizing LLVM, which is traditionally a single-thread code generator.
You said they parallelized LLVM with novel techniques, but then claimed the novelty is the very act of parallelizing it. That's circular, you still haven't explained what are the novel techniques?
Here is a presentation talking about the parallelization techniques, I skipped to the relevant part but the entire presentation is helpful in understanding how Mojo's compiler works: https://youtu.be/yuSBEXkjfEA?t=813&si=MlB7IsBDsj3o-PTX
From experience, Zig compile times are at least as fast as Go for debug builds that don't need to do linking with C libraries, etc. It does have to use LLVM for release builds however which is a lot slower.
Mojo compiles with O2 by default, it is similar to Go here. Release build is the default, debug is something you opt into. It is still very fast nonetheless
Whoever behind the company communication have done an incredibly terrible job. They start with an extremely unnecessary polished Steve-Jobs-alike debut video when they have zero stuff to offer, while the founder name is already immensely trustworthy. Then follow with a very ugly AI-image-heavy website and blog posts, only to be reverted back to plain website later.
I'd love the language and platform be succesful, nevertheless.
It's supposed to be (currently) a fully native, Python-like language, but for GPUs primarily. Though I am seeing it as having a future as a fully native Python alternative that might get interesting.
I hope with 1.0 (havent fully read the article yet) they stop breaking language features / syntax because they did quite a few different changes over time, their overall goal is to be a fully native superset of Python.
I do see this one overtaking Nim over time, which kind of saddens me, because I like Nim for what it is, but Mojo trying to be a superset of Python is a no-brainer for most Python developers. Reminds me of how a lot of Go devs were former Python devs not C++.
> their overall goal is to be a fully native superset of Python
Not any more. That was the initial announcement, but I think then somebody actually looked at Python and realised what a complex language it is - so they refocused on being a Python-like language. Python has a reputation for being a simple language, but in reality it isn’t at all.
Honestly that doesn't really feel like a glowing endorsement of the language. "They announced a crazy goal, then rolled it back because they grossly underestimated how hard it was" doesn't give me confidence in either the long-term design or the people making it, and I feel like "a Python-like language" is both a lot less interesting and a lot more subjective. If the discussions in this thread around significant whitespace aren't already an indication, which parts of Python people think are good and which aren't will vary by person, so without saying more it's hard to tell whether it's a a language someone will love or hate.
To preempt all of the inevitable responses: yes, I'm aware that this was started by the same guy who started LLVM and Swift. That doesn't automatically mean anything he comes up with will be great, especially when it's specifically trying to be "like" a language that doesn't really have much on common with either of those; there are plenty of instances of very smart people having an overabundance of confidence when jumping into something they assume they'll excel at and getting humbled. I'd also argue both Swift and LLVM have some common criticisms that very much are in line with the same concern I've expressed here (i.e. long term design suffering due to expanded or pivoted scope; even the name LLVM alludes to some of that history of having a somewhat different goal originally). His name is enough that I'll probably always click to take a first look at anything he comes up with, but I'm not going to ignore things that seem like issues because of that either. Smart people don't need to held to a lower standard because their work should be able to speak for itself.
Example, Swift for Tensorflow did not end up great.
That is a mistake. Being fully compatible with Python was a big selling point. Now they are just another knock off that wont get enough traction to make a dent. Unless this GPU programming thing is so amazing it takes over the industry with one small niche feature, guess that is the hope.
Language-feature-wise it seems like Nim will be competitive. Ecosystem-wise… yeah it's kind of a nonstarter over in Nim land. There are not many go-to libraries and its culture is not very collaborative. People seem more enamored at what they can hack together than what they can contribute to.
> its culture is not very collaborative.
Second this. Their BDFL's personality is not suited for the job IMHO.
Example: Nim's hash table data structures (Python dict, Go map) are called tables. There are several variants of these, including OrderedTable. Deleting a key from an OrderedTable had O(n) performance because Nim built an entirely new OrderedTable, filtering out the key to be deleted. There were a couple of reasons for this:
- the internal list that preserved element order was only forward threaded, making an O(1) delete impossible
- Nim tables allowed the same key to be inserted multiple times, each with different values. I thought this was a bit crazy compared to other languages.
I tried to changed OrderedTable to use a doubly-linked list to allow O(1) deletes, and to get the multiple key feature removed.
What I didn't realize is that Araq, the BDFL, used ordered tables a lot in the Nim compiler, used the multiple value feature, and didn't want to add the memory overhead of a 2nd list to OrderedTable. His main reason was "deletes don't happen too often". At that point it became impossible to convince him that for a hash table to have O(n) delete performance was ridiculous and would be unexpected for anyone using OrderedTable. I gave up, left, and haven't been back.
Andreas is simultaneously brilliant and myopic to needs outside of his own (compiler dev).
It's amazing how many fights he gets into with Status-IM engineers who are the biggest financial supporters of the project and the only noteworthy company that heavily uses Nim.
Has OrderedTable been ported to Nimony? Does it still have this bug?
OrderedTable in Nim appears to be the same and still have O(n) performance. On a delete, an entirely new table is constructed.
https://github.com/nim-lang/Nim/blob/devel/lib/pure/collecti...
I didn't know about Nimony and just took a look.
https://github.com/nim-lang/nimony/blob/master/lib/std/table...
OrderedTable is now a synonym for Table, ie, both are ordered. To delete a key, the key is looked up to get an index position x in a Seq, the Seq entries after x are shifted left, the Seq is trimmed of the last element, and then the entire table is rehashed because shifting the Seq invalidates the previous hash codes. The performance is better than Nim OrderedTables because at least an entirely new table isn't created, and it probably works fine for small tables, but it still isn't O(1). The performance would be similar to removing an item from a single-threaded sorted list: most painful to remove the 1st element, least painful to remove the last element, but then you still have to rehash the entire table.
Part of that is because there isn’t yet a package format and a central repository. I have a few libraries I would like to contribute or merge with similar projects.
For sure, those things would go a distance towards making it a more competitive language.
Why haven't they happened yet though...? Nim is not a young language. Leadership [Andreas] doesn't think they are more important than spending time adding another memory management option.
I don't know what @winding means, exactly, but at best this seems highly misleading. Nim had babel packages by around 2010 (when it was called Nimrod and there was a Tower of Babel name scheme) and then nimble packages since 2014 or so. Also, literally 30 seconds on https://nim-lang.org (click on Documentation) and you get to https://nimble.directory/ , search for index and get to adix which has all sorts of efficient Table variants.
There has never been much prog.lang. benefit from being "in the stdlib/core" in Nim (unlike Python or Go, say). The benefit is more "software distribution" for the dependency allergic, but the Nim culture is much less micro-deps than seems in vogue lately. All that said, I think having a more batteries core distribution is valuable - just less than you might guess - and practically that needs delegation.
> Literally 30 seconds
And knowing the magic keyword search `adix` to arrive at a prerelease library that you yourself wrote. That not's exactly an endorsement for discoverability. Unless I'm missing something, Nimble is essentially stateless. You have no sense of how many times or how recently a library has been downloaded. These are important proxies.
> There has never been much prog.lang. benefit from being "in the stdlib/core" in Nim (unlike Python or Go, say) … more batteries core distribution is valuable - just less than you might guess
You've been around the community a lot longer than I have. Don't you find this strange? What top 10 language has as weak of a std lib as Nim? [Status all but says](https://status-im.github.io/nim-style-guide/libraries.std.ht...) don't use it.
> practically that needs delegation.
Yeah, I don't think this is the community's strong suit. It strikes me that Araq is tired of explaining himself (but doesn't care to consolidate discussion into ADRs), so when a well-meaning, would be contributor like the parent-poster comes along they get shot down and we lose them.
I only said anything at all because the implication was not "missing minor nits about whatever your favorite proxy for 'will rely' is" but rather "nothing at all". FWIW, the keyword was "index" and the result was adix, and I only meant the package system & "a" directory were discoverable. Exact keyword search systems indeed make discovery harder, but that is some whole other complaint. There are always many complaints.
There are a million ways to define both "top 10 language" and "weak std lib", but the C stdlib is not great and C++'s was bad enough to spawn Boost. And I'm sure many firms all but say "do not use" parts of many top10 stdlibs. In any event, we don't disagree that Nim's stdlib being stronger would be better or that poor delegation or unconsolidated discussion are problems. We disagree on very little, including I'm sure that there is more to a PLang than its stdlib.
What is present/missing in any stdlib (or really in almost anything period) is also often much more subtle and subjective than simple sales pitches one hears. To be concrete, there are "useful" (to someone) things in the Nim stdlib not in either Python's or Go's like std/critbits, std/pegs, ropes, packedsets, intsets, editdistance, etc. Nim stdlib substring search behind "xyz".find() is layered to let you, if you want, pre-build a `SkipTable` (the way regex engines let you pre-"compile" regexes). Subtle in diversity of both kind and granularity and subjective as in "who cares?"
Yes, all that and more is all available in all the ecosystems of anything "popular", but that brings you back to "trust proxies", like "prerelease" numbers, a very weak one, IMO. E.g., I have never used a neovim with a version >0.13, but it's been a trooper; one man's 0.7 is another's 7.0. TRUST IS TRICKY! A count of distinct reliers for some values of "distinct" and "rely" or various update patterns would be better (have their own issues, of course, but at least measures "company" as in "what misery loves", LOL).
You'd likely enjoy: https://mojolang.org/nightly/docs/vision/ for why we created it, and there's https://mojolang.org/nightly/docs/faq/ as well :)
Feedback welcome as always and thanks for checking it out! Encourage you to watch ModCon next week for some exciting announcements :)
> You'd likely enjoy: https://mojolang.org/nightly/docs/vision/ for why we created it,
Ok, yes, this is a substantially better entry point to the language than anything I found by clicking around - maybe link it prominently from the main page?
Updated -> https://mojolang.org/.
"Why Mojo?" is now smack bang on the homepage!
Do you mean the Modular site or the Mojo site? https://mojolang.org/ makes some clear statements.
I was on both. I honestly don't think mojo.lang.org is much clearer - the front page says "GPU kernels + Python + metaprogramming", and the last of those is something Python already has out of the box. The QuickStart doesn't seem to introduce any novel features of the language either.
To be fair, a good bit of the apps/programs/projects posted on hacker news could use a good '1-pager overview'. Like has nobody every explained the 'elevator pitch'. A little marketing doesn't mean you are selling out. Obfuscating your product doesn't make you hip.
This page really helped me to put it into perspective against Python:
https://www.modular.com/blog/an-easy-introduction-to-mojo-fo...
https://github.com/py2many/static-python-skill
You can stick to python and generate mojo or rust or lean.
There is more than one solution to the two language problem.
I think they wrote an internal memo and by mistake they share with the public. I read because one person I know works there, now I don't know what my friend does there either. Sounds like another VC eating company.
For me as a follower, my interest is all about a compiled and fast language with little complexity (vs Rust), and a highly readable and fluent syntax (vs Go).
i think the main idea is that by using MLIR[1] instead of the usual LLVM path it can target gpus directly
Python, but good.
So, without significant indentation?
2026 and still talking about python indentation? what's next, `goto` gripes?
Whenever people stop creating new language with significant indentation. I don't complain about it in Python but I hate when this bad idea is introduced in more places.
its not becauae its 2026 that stupid indentation got better
Oh nooooo, they fucked it up too. Why would someone make a language that’s whitespace sensitive? It’s confounding.
Why would someone make a language where the programmer and the interpreter have two different ways of figuring out where blocks start and end?
at the closing curly?
Programmers look at the indentation.
this discussion is so dated...
llms prefer indents too
They seem to be quite okay with oneliners, have you not noticed ?
To weed out programmers who don't indent.
identation should be something the compiler does not care about
We write for humans, not for compilers.
then let humans do linting the way they like. there is not one kind of human out there.
I never understood this.
What kind of special whitespace requirements do people have?
I like significant invitation, but I still wouldn’t choose it if I was designing a language today.
Functional-style programming has become much more popular since Python was designed in the 1990s, and there doesn’t seem to be a good way to have Python-like significant indentation and also support true anonymous functions.
True anonymous functions has nothing to do with indentation, ML derived languages are also indentation based.
Guido doesn't want them in the language, as simple as that.
Haskell has significant indentation and anonymous functions.
Haskell's significant indentation is, in my experience, much worse than Python's at fading into the background by matching what you would have done anyway.
As I understand it, Mojo’s goal is to be able to match state of the art performance for parallel algorithms and be able to use the exotic GPU hardware. A functional style is not necessarily productive in that context.
One interesting PL concept they implement are linear types https://mojolang.org/docs/std/traits/anytype/AnyType/
Scala 3, F#, CoffeeScript, Nim, Haskell, Elm, PureScript...
Like with array base indexes, there are plenty of languages that start with something other than zero, and many that use indentation, including a few well known in hacker circles, Python isn't alone.
It's really not that bad
When has it ever been an issue in use?
People complained about this in 1999 but then I used it and it's never mattered.
It’s an issue when generating python, I can tell you from experience. It’s also a pain to not be able to safely rely on an auto formatter
Every python developer I know uses an auto formatter today, and never needs to think about it (ruff, black). It's not ambiguous and is immediately visible if something has the wrong indent.
(On other projects, I often wish other languages had formatters as good and standard as the Python ecosystem. I have many complaints about Python but whitespace is just not a problem I've seen)
Today and forever, I was using ActiveState IDEs back in the day, I started using Python in version 1.6.
Also it isn't the only one, many ML derived languages also use indentation, including the famous F#, Standard ML, OCaml, Haskell,....
You missed the point. I’m well aware python has formatters, I use them and implemented my own. The point was that when generating/manipulating python code you cannot just generate in whatever format then run a formatter. you need to carefully keep track of the exact indentation at place where the code in written
Fair. That is a very slight friction on what is quite a small niche, though (compared to the wider ecosystem of developers). And most that need similar will use AST or runtime construction.
The trivial kind of raw source generation is still trivial, but I'm guessing you're doing embedding/templates or something?
It’s definitely niche!
If I were doing this, and the code generator was itself in Python, I'd probably use ast and/or tokenize to avoid having to manipulate literal whitespace.
Python's significant indentation does have real costs, but they mostly don't affect experienced programmers working on real codebases in real editors or IDEs. Rather, they are things like, it makes the REPL more cumbersome, and newbies sometimes find that code they copy-paste from the Web mysteriously fails to work.
Have you never tried to paste python code from the internet into an editor? At least half the time the indentation is screwed up and I have to go in and figure out manually where to put the spaces in. Indentation gets screwed up by pasting in other languages too, but with those I can generally tell the editor to reindent and it can because the information is included in the source code.
Get a better editor! Yes if your tools suck, things suck. So get better tools.
I've tried them all - vim, emacs, vscode, etc. They all fail in this regard not because of themselves but because significant whitespace is a terrible idea which nearly every programming language avoids.
the solution is the same - you tell the editor to paste with an indentation adjustment
Significant indentation is why people love python. Can 2 billion people be wrong?
yes
I could get started about religions now...
yes the majority is often very wrong
I mean, it feels like "Python, but aiming at a completely different vertical that values native compilation and a borrow checker"? Which is not quite the same thing
Don't see the value of using a language with a closed source compiler...
Much better options out there. Python already has libraries like Pydantic that offload performance to functions written in Rust under the hood.
> Finally, we will continue to progressively open-source more of the Mojo language, as well as components in MAX that we have built with it. Our commitment remains unchanged – we will open source the Mojo compiler and toolchain in 2026.
It's categorically uninteresting until and unless that happens. Weird to release 1.0 without it. What are they waiting for?
> Weird to release 1.0 without it. What are they waiting for?
Hmm, I've been under the impression that at one point they said it'd be 100% FOSS by the time 1.0 hits, but maybe I misunderstand/misremember the details. But I remember they'd said it'd be FOSS in connection with 1.0, seems weird to do it this way instead of waiting until it's all FOSS then cut 1.0.
https://news.ycombinator.com/item?id=48057901#48068126
Make of that what you will.
They have some event scheduled for next week too.
People to tune in to the conference, probably.
This exactly. Doesn't scream confidence that its still closed upon release.
Why? The creators have a strong open source background and it's work to properly open source.
Because they promised 1.0 would be 100% FOSS. That has proven to be a lie.
From listening to Chris Lattner in interviews my guess it’s learning from when Swift became open source. The beta versions of Swift had massive language changes in how value types worked like with Array. Mojo is trying to get this language design locked down first, which is easier to do behind closed doors.
A compiler promising source stability is a different commitment than the compiler’s code being ready for outside contributors. Which is also different to having ABI stability, which took until 5.0 for Swift to achieve (I think Mojo does not have ABI stability as a goal as it doesn’t have to link to OS frameworks).
Massive language changes are easier to do without users. Whether the source code is visible to external people or not is completely irrelevant. I have a few programming language implementations on my GitHub and let me tell you, making manor breaking cha he's to those are easy as pie when I'm the only user.
If Mojo's goal was to make evolution easier, they would have a limited closed beta like what Jai does. But they're not doing that; they make the toolchain available to everyone, they just did this marketing push around releasing version 1.0 of their language. They're making it harder to make breaking changes by seeking users among the general public.
I have no idea why you started talking about outside contributors though, nobody's arguing that they should accept outside contributions. You do t have to do that. It's a completely orthogonal issue to what we're talking about, which is releasing the code under a FOSS license.
That doesn't really make sense, as being open source really has nothing to do with whether the language has an open evolution process. And if they've already declared 1.0, then they've already decided they're stable, so the original point doesn't apply any more.
I feel like they're going to remain closed until Qualcomm is convinced they can't make any money from it being so.
I think they are waiting for a wild success in adoption, and if that doesn't happen they can use the IP to pivot to something else without giving the core away to anyone for free/as open-source.
A lot of people beg to differ. And it's their decision, not yours on when to open source. They have their reasons even if they aren't explaining them to you.
also what license will they be going for?
The point is that Mojo compiles to MLIR for programming on heterogenous compute and GPU so something like Pydantic isn't really comparable here.
The alternatives would be stuff like CUDA
With cuTile Python, allowing to write Python today, without changing languages and first tier support from NVidia.
The value has already been materialized. They were acquired by Qualcomm for $3.9 billion.
The value has been speculated on. It doesn't mean the value has materialized.
Wow, so Lattner became a billionaire. Pays to be in compilers.
For once. A former colleague once quipped: The only way to make decent money in the business of language runtimes is to get a job at a big company that depends on it.
Yeah, feels kind of odd to phrase it like that when he's already created more than one world famous compiler before this one.
TBF, he has done more than his fair share of time working in BigTech.
Sure, but that doesn't mean he made billions before!
That’s insane, for a language that seems to be Temu-Python.
OTOH Python itself is the world’s most popular programming language, yet has only recently had the money to support more than a single full-time developer.
What else are you going to write CUDA kernels in that will be better than Mojo?
This doesn't compete with Python.
Like if someone wrote a competitor to VHDL that looked like TEMU Python... Python would not be its competitor, VHDL would.
Julia is actually quite nice for this. If you prefer a python-like approach consider Triton from openai, numba (https://numba.pydata.org/) or CuTe DSL from Nvidia.
Python tooling from NVidia itself, which generate the same PTX as CUDA C++.
Julia
lol no
NVidia is of another opinion.
https://developer.nvidia.com/blog/cutile-jl-brings-nvidia-cu...
it's legitimately among the best languages for writing GPU kernels
As someone who writes GPU kernels in Julia, yes it is. But Mojo is one of the few alternatives based on technical merit alone. Each language is better in different metrics, and it could be argued either way.
License-wise, Mojo reminds me of paid Borland compilers that were in vogue before I was born. Even if it's open-sourced, what incentive does Qualcomm have to maintain a happy path for Metal deployment? Mojo might actually be the best way to program Apple GPUs (on a technical level) and it would still lose due to politics.
If it's open sourced, Qualcomm won't need to worry about Metal support. If Mojo is popular enough, Apple might get involved. If not, no doubt there'll still be Apple users that want Mojo support, and are willing to work to get it.
Look at the Asahi Linux effort, for instance...
It almost feels like they were paying more in the hopes of owning the next LLVM or Swift based on the name on the tin rather than basing the price on the product itself.
C and C++ were closed source for quite some time, they still are in places like commercial embedded OSes and console dev kits, even when they happen to be GCC and clang forks nowadays.
Clang actually has replaced most GCC forks exactly because vendors can keep the fork closed source.
Sure but at those times C and C++ had no immediate replacement, and if you were doing embedded C/C++ was your only option. The same cannot be said for Mojo.
There were competitors. Pascal and assembly, most obviously.
It was just a different culture. There was no expectation that everything should be free.
In that regard yes, personally I think Mojo has lost its opportunity, but not due being commercial.
AMD/NVIDIA/Intel now having first class support for Python DSLs or GPU JITs for their APIs, and Julia as well.
The C compiler source came with the AT&T UNIX tape ... it wasn't open source because FOSS didn't exist at the time and it was under a proprietary license, but it was not "closed source". Parts of MOJO, OTOH, are closed source because the source has not been released (yet ... it apparently will be later this month).
That one yes, the ones sold by all UNIX clone vendors definitely not.
In fact, GCC was largely ignored until Sun decided to split SunOS into two SKUs.
And then you drag the goalposts out of the country ...
As I said, C was not "closed source", it was distributed with source under a proprietary license. And "all the UNIX clone vendors" didn't exist for years after the UNIX tape was first released. And they weren't "clones", they were enhanced supported versions of the AT&T system. I worked for nearly a decade for the first of those commercial UNIX vendors, ISC ... the other prominent ones were SCO and Wollongong. (Sun came along later, based on BSD UNIX, developed for the VAX and of course derived from AT&T UNIX ... none of these were "clones" ... the first of those was Minix, and then GNU/Linux). We all distributed UNIX with the AT&T C compiler. Eventually ISC wrote their own C compiler for IBM's AIX project because IBM always had special requirements ... I don't know whether we distributed the source or not.
But the point is that this was nothing like the situation with Mojo, which currently has not released the compiler source ... which is nothing like what people on this page are writing, because it will be open source SOON, probably coincident with the conference later this month.
I won't respond further about this.
Pydantic is mainly about making Python strongly typed.
Mojo addresses the actual need to write code in other languages, such as C++ or CUDA, for low level and hardware specific work.
But I wonder if it has already missed the AI hype cycle, when a lot of low level code is being written. I'm not sure how the developer ecosystem will react now.
Their website has a banner suggesting news about open sourcing of the compiler at some Aug 18 convention.
They are going to open source it eventually
AI generated first image does not give me much confidence.
Lastest OpenCV 5 release notes also had a lot of LLMisms.
I guess that's the new normal. Still, I am very hopeful for Mojo.
I know right? Would be better to have a banner with the language logo than some totally out of place aislop image, almost makes you think this is a random person's blog and not the official Mojo site
I believe Mojo has been primarily using AI-generated imagery of its mascot for several years now, which politically aligns with its pivot to being a language for AI development.
I wouldn't say there was a pivot, AI development has explicitly been the goal since launch.
Oh, gotcha. I thought I recalled reading early material that emphasized the flexibility of Mojo emitting MLIR for many tasks, but it's possible that that was either pre-launch, or that I got confused when reading about MLIR itself.
A lot of fancy words to say this is slop.
I was trying hard to be diplomatic. :P
Oh wow a business pivoted to AI what a shocker lmao (not directed at you/mocking you, to be clear)
Almost? That's exactly what it did! At first I thought this was some bozo's toylang mysteriously on the front page of HN garnished with a wall of LLM technobabble.
I spotted the exact same thing with the AI image. makes the whole thing immediately feel cheap, it's wild
They’ve had this branding for some years, at least since 2023. I doubt at that time it was fully AI generated, although I have no proof of it. I always looked at it as being heavily AI inspired.
> makes the whole thing immediately feel cheap, it's wild
Hard to decide if the blog post is wild, or people's visceral reactions to it.
I've said it before: I believe the obsession in trying to figure out if something is AI generated will soon be classified as a mental disorder.
nah, i just have no interest in this endless slop. It's so fucking obvious, so cheap, and soooo disrespectful imo
I think AI use will soon be classified as a health risk worse then smoking, drug and gambling.
There’s no obsession. It’s an obvious, ugly slop.
Ugly images in blog posts were common before LLMs came around. People on HN didn't feel compelled to point it out each time.
"Normal" keeps getting more and more depressing.
Did they think that image would make them look fun or interesting? Because it's exactly the opposite.
These AI illustrations are the nee "corporate Memphis." Images of flat, minimalist people in bold colors used to be the sign that whatever text accompanied them was low-value drivel. Nowadays, I associate the same with AI generated imagery. I'm not sure why the folks doing the writing (if, indeed, any of the writing was done by a human) don't seem to clock this when the audience seems to be able to catch on almost immediately.
Isn't Mojo expressly for building AI frameworks? It isn't really needed or special as a language by-itself. It's main purpose is to make it easier to program these high end graphics cards for AI.
So maybe an AI image isn't the worst.
People who work with AI all the time should be the most finely tuned in to how sloppy it looks and when to use it and when not.
A lot of software engineers don't have a good feel for this kind of aesthetics or social outcomes. They're playing with their new toy (which is very exciting, let's not ignore that), they're happy it does stuff, but ignore that it can be perceived as ugly and cheap by others. As well as totally bland - every other cheap company is using it too.
Yeah, I just assumed that it was some sort of image generation crap >_< It's funny how this affects me tbh, I think it has almost the same effect as if there would be a boner-pills ad there or something. Or a 'subscribe to read' popups.
It's all just spam, filler content and I did not ask for that.
> Lastest OpenCV 5 release notes also had a lot of LLMisms.
Yeah, that really made me triple-check if I visited a fake SEO spam website instead of the official OpenCV one. The fact that their blog looks exactly the same every single low-effort AI content farm doesn't help either.
> I guess that's the new normal.
Yes, so basing confidence on it is totally bogus. Such comments are cheap and add nothing to the discussion of Mojo reaching 1.0 status.
Well, not really; people have long ignored overly corporate or low-effort news releases because. The presence of certain language or stock images is a signal (and I would argue a valid one) that the rest of the content is equally low-effort and not worthy of attention. AI images are just the newest version of that signal. I think it behooves the writers of these blog posts/news releases/community updates/etc. to reflect on why they continue to include these signals that their content is low-effort or low-value if they actually want their audience to read it.
Unless the idea is that, in 2026, you assume your audience is an LLM that is ingesting and summarizing your content for the end user and that no human will actually read what you wrote so you don't actually need to care about signals of quality or trustworthiness. The AI generated illustration seems kind of superfluous if that's the case, though.
Finding some unit of the thing one can actually understand and using it to issue a shallow dismissal is canonically the most Hacker News possible comment. And the rest of the commenters generally love it.
Not to worry though, if there’s nothing webshit accessible in the content there’s surely something wrong with the CSS to complain about!
Is Mojo being a superset of Python still part of its projected feature set?
I tried Mojo a bit early on and decided to come back to it when it was more complete, and in the interim I read some comments somewhere that the superset-of-Python part was being walked back from.
On https://mojolang.org/docs/roadmap/ it's stated (Phase 3) that "Mojo may or may not evolve into a full superset of Python, and it's okay if it doesn't."
I feel like that "full superset of Python" part was part of the appeal of Mojo.
Personally I'm kind of tired of 97% python-compatible this-and-that. Python itself has enough of a history of incompatibility and the cognitive overhead in switching might be easier with just moving to a totally different language. But maybe not — it depends on the details I guess.
Honestly reading through the documentation more and more I'm more and more confused about what the 1.0 release actually means.
> I feel like that "full superset of Python" part was part of the appeal of Mojo.
It absolutely was, but IMO was never going to happen. Python has a reputation as a simple language, but it really isn’t. It’s a very complex language hiding behind friendly syntax.
I think once the Mojo team realised that, they had to adjust their expectations.
I know Python since version 1.6, the reputation I guess is mostly because a large majority doesn't read the well written documentation, and mostly use beginner level constructs.
Python is C++ level complexity, when one fully understands the runtime, its dynamic capabilities, the C API, the standard library, the OOP model reboot on 2.0 with new style classes, all breaking changes that happen between releases, and naturally one doesn't know before hand what are the exact requirements to execute any random Python script.
They moved the goalposts. I think the landscape has changed since their initial Python superset pitch.
It seems like a deliberate pivot towards creating an AI-era language which make GPU programming as easy as writing Python.
To be fair, translation plays to the strengths of AI, so there's less benefit to being an exact superset.
> Finally, we will continue to progressively open-source more of the Mojo language, as well as components in MAX that we have built with it. Our commitment remains unchanged – we will open source the Mojo compiler and toolchain in 2026.
Why not now though? Why wait for 4 more months? What magic is there to hide in just making it source available at least.
It's probably very messy and not a great baseline for outside contributors to start create PRs.
That's the favorable interpretation, atleast.
It's likely AI slop.
I made a language design some years ago. More recently, I had Claude carry out some of the implementation to see if it would work. It did not. I tried again, hand-tailoring the IRs and many of the main algorithms, and it did better but the self-hosting compiler still violates many of the design principles and even behaviors in the design document. I have spent six months of free time reviewing and fixing it myself, and it's just a small pet project. I would not open-source that code any time soon.
I must imagine Mojo is in the same situation.
Ladybird is Open Source and doesn't accept outside contributions any more same with sqlite and many other projects. Lots of Google's oss projects also are not very open to contribution.
At this point I find it deeply troubling that it isn't open source. I respect Chris I know folks who have worked with him and but at the end of the day I feel too weird about this entire closed source programming language thing.
If we complain to Nvidia about there weird closed source limitations, I think Mojo is just as bad. You don't even know what your compiler might be doing, and if it does something wrong you can't fix it, you now have to file a bug and wait until the gods from above come to your rescue.
I have had similar issues with drivers, I have faced soo many driver bugs in my work and the answer is always write more code that works around it somehow, this warp level is bad, this feature doesn't work with these values, etc. etc. Why can't I fix them, I have often on Linux just pinged the Mesa team about driver bugs with a rough draft of a patch and gotten it fixed, I can even patch and run it myself when I am in a hurry.
I feel like atleast making stuff, source available like Unreal should be the baseline for any "core" software where a bug could possibly be a major blocking issue for a developer. Windows and Mac are very bad with this.
Mac's CoreML has had soo many bugs over the last couple years I have had to email someone I know at the company to get them fixed or try to get into contact with someone.
Intel's drivers have had bugs too but I have had more contacts there which is a wash. AMD well it's AMD...
Either way closed software sucks I don't want to use fully closed stuff unless I don't have a choice people can make what they want of it. I don't really care.
I care much less about just foss.
They could go the sqlite route for it then, no outside code but let us have a look
My guess is that the compiler will be opened next week at Modcon. Honestly? I'm surprised Mojo 1.0 is being released now; I had thought it would be released at Modcon along with the compiler source code, but oh well, 18th August isn't far now.
Who said anything about 4 months? Your "logic" seems to be that they are releasing it in 2026 and there are 4 more months in 2026 so they will be releasing it in 4 months. That's obviously flawed thinking. They will probably be releasing it coincident with their August 18 conference.
So why not either delay the 1.0 release by a week or move the open sourcing up by a week? This "strategy" makes absolutely no sense.
From Mojo Wikipedia article: ``` The intention behind Mojo is to bridge the gap between Python’s ease of use and the fast performance required for cutting-edge AI applications. ```
Basically, same raison d'etre as that of Julia.
Completely true, but Mojo approaches the problem from a different angle. Where Julia aims to blend fast and slow code seamlessly, Mojo's solution is to have a set of opt-in semantics that ensure fast code.
Mojo and Julia have some very fundamental differences though...
Mojo is a fully ahead-of-time (AOT) compiled systems language. Julia is much more dynamic, and its AOT compilation story is still quite weak. Mojo also has explicit memory management, unlike Julia and its real time unfriendly garbage collector.
Mojo also has the concept of ownership, similar to Rust, while Julia does not.
Mojo is being pitched as an efficient, general purpose, Pythonic systems language which can target a variety of hardware types. That seems like a good target to me, I hope it's successful!
One thing to keep an eye on is whether or not a Windows port emerges after it's open sourced...
The design and intellectual heritage of Julia and Mojo couldn't be more different though. Mojo is a Swift/Rust style language from industry veterans, Julia is a Lisp built to make academics write faster code.
Nothing but respect for Chris, and I understand why founders take these deals. But an acquihire usually means the product is the part that gets left behind, and that's the part I cared about.
$3.9 billion is not an acquihire!
Well, there was $300M+ in VC money and after burning through some percentage of it those investors wanted an exit. I don’t think this is so much about founders taking a deal as it is investors demanding that exit.
Apparently they'll be open sourcing it on August 18th, one week from now.
So is the language itself proprietary licensed, with standard library Apache 2?
Am I understanding the current state of things correct?
What in the world would posses the modular team to attach those horrendous AI generated thumbnails to each blog post? It makes it feel cheap, even though I know the project has some really serious technical expertise behind it.
They're an AI company
AI infrastructure. Regardless, I don't see OpenAI or Anthropic attaching AI generated images to their blog posts.
Ssssh, don't tell them.
"AI slop image below the headline" is an unmistakable mark of low-quality content written by a sophomoric AI evangelist. Let them keep watermarking their own content.
See their in-depth vision doc here:
Mojo may be interesting to me, but they should really split it from Max, which is some closed stuff I'm not interested in.
They should show some performance comparisions between PyTorch and Mojo, PyTorch+kernel compilation + Triton vs Mojo, ThunderKittens vs Mojo.
> They should show some performance comparisions between PyTorch and Mojo, PyTorch+kernel compilation + Triton vs Mojo, ThunderKittens vs Mojo.
The fact that they release 1.0 without doing this, I think says a lot. I personally haven't even started looking into Mojo because of the closed source stuff, but usually you can tell what's going on by looking for what's obviously missing.
They kind of did. It has a separate website now [1] . All the docs, announcements, roadmap, some comparisons as you asked are there. So they do seem to be working towards separation.
Thanks, it's so boring and disappointing.
LLVM has such a cool tutorial, eventhough I haven't written a programming language myself, it was excitung to go through it to see how static programming languages are lowered to SSA form.
Mojo should have something showing how to write a super fast matmul and especially what's interesting for me is a fast linear attention kernel, as that's where PyTorch is getting much harder to use.
Aren’t they already separated? Max is an inference runtime, written in Mojolang, which uses the MLIR compiler chain to target diverse accelerator hardware.
This probably has a simple answer, and is something I can easily google. But it's always been in the back of my mind, what optimizations would be possible if you took actual python code and forced 100% type annotation. Could you meaningfully speed that up at the interpreter or with some pre-processing or compilation?
EDIT: I guess this kind of answers my curiosity: https://programming-journal.org/2023/7/2/
Python to C++ compiler: https://shedskin.github.io/
Compare the same algorithm/program in Python and C and that's roughly your answer, no?
Given the shared heritage with MLIR, one way to think about mojo: higher level IR, but still an IR. Rpython for GPUs.
But mojo is a superset, not a subset. So why not use a subset, infer what you need and generate mojo?
I've never seen this question actually presented to the company and discussed in more detail.
never has their been a language release so uninspiring
A bit odd to have a 1.0 release without Windows support.
The main downside is that usually I only choose languages that the LLM is extremely good at. (however I avoid dynamically typed).
But I hope they succeed, it's a very interesting approach. Then I will jump on it later on. S
The premise of Mojo used to be that it was a superset of Python so in theory LLMs would benefit from that but I'm not sure how much divergence there's been
Wouldn't that cause an awful lot of confusion in practice? Instead of having one being a superset of the other, you now have two almost identical syntaxes, but probably not quite.
LLMs have enough problems distinguishing between major versions between packages, e.g. where all the imports were moved around and renamed.
Are you aware of this?
Are there any benchmarks on which languages models perform best?
Sorry Chris, love you but will not use a closed language.
They're open sourcing it next week. It's in TFA
TFA says "we will open source the Mojo compiler and toolchain in 2026" and "we’ll share more on our plans for Mojo, MAX, and open source at ModCon on August 18th" which makes it seem they're not open sourcing it next week, but some time between next week and 1/1/27.
Interesting. Heard about Mojo a few years now, never really tried it. My Python brain cannot really write C++ or Rust but still need the performance.
You may have heard of it already, but Nim has existed in this space for some time.
Nim could do better with more than just 1 or 2 brilliant developers
Julia as well.
Take a look into numba, you add a decorator to your function and it turns it into compiled code.
It's JIT only, and it needs to be in a particular shape to be effectively compiled. It's mostly for numeric code, but there are many other cases where native code would be better.
a systems language that reads like python is great!
python superset would have been perfect.
I do hope the language thrives and gets a community.
All the best to the modular team.
> following the standards of how mature languages (e.g. C++) evolve over time
This gave me a chuckle. When I think of languages evolving their standards, I do think of C++. But not, you know, in a good way.
The point is that you can compile both old and new versions of C++ using the same compiler. Rust and Go also evolve like this
You may compile, linking binary libraries compiled with different versions is a different matter, especially if there are relevant changes between versions on either the standard library, or other dependencies.
> uv pip install --upgrade mojo
What's wrong with people?
I wish Modular/Qualcomm good luck with this. We would be well served with a good alternative to Cuda. I will be looking at Modcon to see how serious they are with Mojo's ecosystem. (Hopefully it won't be yet another SYCL).
first impression based on the hero image is that this is a new model for image or video generation!
Hurray!
the 1.0 release of Mojo, it's a pretty big event, of course first thing I'm greeted with is an AI slop image on the top of the blog post
awesome
like finding a rock of gold underground and tossing it away because it’s dirty?
Clear, useful, done
"We arbitrarily decided we are at 1.0 now, this is a major milestone."
closed source ngmi
Creating a new language in the age of AI seems kind of pointless. I don't write code by hand anymore, I only review it and I prefer it to be in a language which I already know since investing time in learning another language doesn't make sense at this point.
Pointless? If compiled mojo is faster, more memory-safe, and more hardware independent, that doesn’t seem pointless at all to me.
(And now in the age of LLMs, the barriers to get started with a new programming language are lower than ever)
Eventually we will all be writing COBOL and how English turns into machine code is irrelevant.
Do you know CUDA or Triton? The point of Mojo was to replace GPU kernel programming langs with a Pythonic one that was more familiar
And Mojo will die because nvidia is not interested in maintaining any semblance of backwards compatibility when it comes to CUDA. It's going to be a perpetual race forwards, and whenever nvidia upgrades CUDA, you're going to be stuck there twiddling your thumbs and waiting for Mojo to become compatible.
Additionally, anyone who has seen how Swift evolved should be wary of Lattner saying "the language is mostly stable", or even "the language is released".
It's not true that Nvidia doesn't care about CUDA backwards compatibility. The invention of TileIR was motivated entirely by that problem. GPU architectures are supported by the full tool suites for at least 10 years by CUDA (that's longer than they are for graphics).
Nvidia is also a major contributor to MLIR, which they use for TileIR and ClangIR, so should a compatibility break occur it wouldn't be entirely up to Mojo to fix it.
None of these things are true except the ClangIR thing
this language predates the rise of "AI" (you probably mean harnessed LLMs, but I'll forgive you) by several years. This is the 1.0 release.
This is the team behind MLIR, they have been in this field for 10 years or so.
Does it? Mojo was announced 6 months after ChatGPT was released.
Sounds like you're just lazy. Which is fine. You'll probably have a okay career. But not a great one.