I haven't seen anywhere claiming they are open weight (although their last similar model, NLLB was).
They say their leaderboard and evaluation datasets are freely available. Closest statement I've seen in the paper, "Our translation models are built on top of
freely available models."
Meta released No Language Left Behind (NLLB) [1], I think in 2022. I wonder why this in not "NLLB 2.0"? These companies love introducing new names to confuse things
This project is absolutely NLLB 2.0 in spirit. However, we decided to reserve the name “OMT-NLLB” only to the subset of the new models that have encoder-decoder architecture similar to the original NLLB-200. The other models are called “OMT-LLaMA” and have classical LLM architecture.
The idea here (and we had to emphasize it to justify the project internally) is that we are developing not just new models but a recipe for massive multilinguality that can be integrated into general-purpose LLMs.
I'll be looking at this in detail. I've started a company to do similar things, https://6k.ai
I'm currently concentrating on better data gathering for low-resource languages.
When you look in detail at data like Common Crawl, finepdfs, and fineweb, (1) they are really lacking quality data sources if you know where to look, and (2) the sources they have are not processed "finely" enough (e.g. finepdfs classify each page of PDF as having a specific language, where-as many language learning sources have language pairs, etc.
Hey, this is super cool! I’ve been working on a similar problem, focusing on low-resource and underserved languages including the Mayan family, and have published some research and open resources around that [0, 1].
On the data side, I’ve found that the biggest bottleneck isn’t collecting text (it’s out there!) but reliable language identification. It’s often difficult or ambiguous to separate languages cleanly in datasets like Common Crawl, Fineweb, or others. I worked on improving this a bit for Fineweb 2 for my native language, that might inspire you [3].
Many of the challenges you mention seem to recur across regions and language families, so I’d love to connect and compare notes sometime. Feel free to reach me at omar [at] the labs site below.
Excellent, thank you mandeepj! Curious about the language coverage of your agent and if / how you plan to eval your agent, if you're willing to share more.
It’s a small sample and not specifically ones we’re working on. It’s biased towards alternative scripts for visual interest.
Australian languages are definitely interesting! and I will say, from what I’ve seen, Australian government (and other orgs) have done better than most to help document them (in recent years, at least)
Yes, there are government datasets, languge "acadamies" (or "regulators") - organizations focused on preserving / teaching the language, and often smaller, local publishers that publish material in their local language.
I'm living in Guatemala, so have been focusing on the Mayan languages here (22 languages, millions of speakers).
As an aside, I remember visiting Guatemala (in the border area near Chiapas) in the early 90s and discovering that “Mayan” was not the monolith that I had been led to believe by my culturally narrow American education, but was a diverse collection of related cultures with multiple languages.
In one of the villages we visited, there was a language school where foreigners were learning Jacalteco. One student was from Israel and where most of the students had vocabulary lists in three columns (Jacalteco - Spanish - English), his had four columns where he did one more step of translation to Hebrew.
So, LLMs are noticeably better in Khmer than Google Translate? I wonder why Google Translate doesn't use Gemini under-the-hood. Perhaps it's more prone to hallucinations.
I'm interested in find some thorough testing of translations on different LLMs vs Translation APIs.
There's a dropdown on Google Translate that lets you choose "Advanced" mode or "Classic" mode. Advanced mode uses Gemini but it's only available for select languages.
Recently asked Codex (GPT-5.2) to write a small single-page HTML frontend to debug some REST endpoints. As it was just a one-off tool, I put in no instructions about looks or styling at all. Lo and behold, the tool it wrote came with exactly that round-box style.
It seems to be the "default" style of some models for some reason.
Which makes me wonder if people already experimented with different style suggestions to get different results: "Make it look like an 1998 GeoCities page" / 2005 Facebook / Newgrounds / DeviantArt / HN / one of those Windows XP simulators with built-in window manager / etc
I vibe code web apps with Google's Gemini and I think it actually mimics Google's UI and UX because I see similarities between my vibe coded web apps and Google's web apps.
Every vibecoded site have this same dark look with shining hue-gradient borders, can't wait for the future the entire web be filled with this generic look
This is fair, although I ask for it to be dark themed to match what I think was the style of typing game I remember growing up with (it's been a while). Bumped up the font though.
Next time please ask it to respect system dark/light mode preference, it's trivial to do, especially for an LLM which can spin up light/dark alternatives easily.
By "free windows" do you just mean an unactivated copy of Windows? That doesn't prevent the user from configuring their preference in the browser itself.
My top complaint is that if I've successfully used a pattern, I want my text removed. I keep forgetting to backspace a bunch, then get frustrated that my pattern isn't working.
Automated accessibility testing needs to be in your loop, whether you are using an llm or not. Aria labels are easy to get right but they are also easy to forget.
There's a number of recent, good quality, small TTS models.
If the author doesn't describe some detail about the data, training, or a novel architecture, etc, I only assume they just took another one, do a little finetuning, and repackage as a new product.
reply