embedding-shape
Born on October 27, 2025•15690 Karma
Does multiple vendors run this "single API" or how is this not replacing a single-vendor dependency for another single-vendor dependency?
> Fully open model: open weights + open data + full training details including all data and training recipes
There are equally open, much more useful models out there: https://artificialanalysis.ai/?models=nvidia-nemotron-3-ultr...
That it's language agnostic and somehow matters feels weird now a lot of time (and experience I suppose) afterwards, but back when I only knew 1-2 languages by heart, also having to face understanding Python at the same time from Norvigs guide/reference made it slightly more complicated for me.
I use this as a litmus test now when coming across new languages (implementing MAL in the new language), as it's such an easy approach to practically test large parts of the new language, and there is always host-language-specific tricks you can learn along the way.
plusnirreg.com (misspelling intentional) is the home page of the license
It's also AGPL compatible!
> Why?
By including the word "NXXXXXX" in a LICENSE file that must be distributed with the software you will ensure:
The software will not be used or hosted by western corporations that promote censorship
The software will not be used or hosted by compromised individuals that promote censorship
Users of the software will be immune to attacks that would result in censorship of others
With the exception of DwarfStar + DS4-Flash with IQ2_XXS quantization, which somehow seems to not suffer as much as I'd thought. I'd still opt for a smaller model + at least Q8.
All these groups of people sometimes play in the same lobbies, and what the players "gain" from the session can be very different depending on the person. There is no "right or wrong" way to play video games, or the right/wrong motivation for it, it's just different.
I think they might destroy the streaming/online communities, but I wouldn't say it destroys the game itself. I play BAR, but never with random strangers, the game works fine, but I also don't participate in any "video game" communities or watch/play with streamers, so what you're saying sounds very foreign to me, and is more about the communities than the games themselves.
I only play public matches with random strangers and this is the feeling.
Obviously this wouldn't apply if I had a small community of not-strangers to play with consistently, which you do have but oddly describe as not having a community.
If what GP is saying is "play with people you know personally and then you won't have to play with people you don't know personally," well, sure. Great insight.
Most people don't and can't do that. That's why online matchmaking exists and constitutes 99.999999% of online gameplay.
Ideally, it should allow non-competitive players of similar performance level to play against each other.
Which think about what that feels like: getting semi-consistently beaten by worse players who just all "happen to have" the exact same loadout and exact same strategies and exact same everything.
That's exactly what I'm describing. It's incredibly boring.
Losses against someone with better logistical choices is normal and expected. You should quickly pick up on what they were doing and learn to counter it. If you can't then how can you claim to be better than they are?
I recognize the adjacent commenter's point about the small population though. It might be difficult to be appropriately matched up unlike a AAA title at launch.
> You should quickly pick up on what they were doing and learn to counter it. If you can't then how can you claim to be better than they are?
You must not be familiar with what "meta" means. Modern video games seem to all have an across-the-board superior configuration. And yes, the way to "counter it" is to copy it, which is why this strategy spreads like a virus through the game until 100% of the people who beat you are all doing the exact same thing.
If the desired outcome of the game design was for everyone to use the same configurations (the meta) in order to "counter" the meta, then why have configuration options at all? Give everyone the same exact thing all the time.
The reason they don't do that is because it's extremely boring. But unfortunately the "meta-finding" capability of the streaming community yields the same functional outcome anyway. Ergo: the games are made boring by the community, as stated at the very top.
And no, the issue isn't "I like to casually play with off-the-wall builds." The issue is "video games were a lot more fun when you encountered different types of opponents."
This is, of course, why game designers put so much work into supporting variations in builds, so obviously they agree too.
I didn't criticize anyone for being less skilled or anyone for being "in the wrong." I'm observing a game dynamic that makes games less variable than their designers clearly intend.
If you don't like it that's fine - I'm actually mostly in agreement with you. But I don't think it's necessarily a flaw in the game any more than chess openers are necessarily a flaw with chess (although chess960 does exist so clearly not everyone appreciates the typical openers).
I'm not saying people who do this are evil or bad. I'm saying they make games not fun, and it's not clear at all that this is solvable by the game designers except by vastly simplifying the games (e.g. getting rid of builds).
The fact they design for variance so much but then do not achieve variance is an empirical proof that it is a flaw in the design.
Yes, this is also a flaw with chess and is why chess is an extremely boring game. You can just ask someone their Elo and know who will win. As you get into higher Elos, variation in play style becomes less relevant, i.e. more homogenous. This makes the game boring.
It's not my job to lend airtime to other people's opinions on the matter. I'm sharing my opinion. You can share yours if you want.
I've used this a bunch as a suffix to try to prevent that, works OK in most cases, but not always obviously, works better in the system/developer prompt if you have access to those. Seems I've used that about ~1000 times since 2025/08 when I started using codex (- transcription duplications, so maybe 1/2 of that?).
$ rg -a -o "Answer grounded in truth" ~/.codex/sessions | wc -l
1046Indeed, it's easy to surface this by sending one model a "Review" of their proposal to another, then bounce them back and forward, ask which one is best and both models will almost always say something like "The other proposal/review is better", I'm guessing because somehow they think it comes from the human, and "human is always right" or something.