Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Muse Glimmer: 30B-parameter model optimized for always-on local agent workflows

research.meta.ai

Post reply on HN
Print view View original

291–300 of 682 posts

Re: Muse Glimmer: 30B-parameter model optimized for always-on local agent workflows

#291

Post by combilabs »

ericmay wrote: post #171
avaer wrote: post #101

I lament the comments saying this in any way redeems Meta (the company). The researchers releasing this stuff have almost nothing to do with Meta other than being bankrolled by the slaughterhouse. You aren't the customer, you are the pawn in big tech's game of thrones. Your good will is a commodity to be traded, almost literally. It will be used against you the moment it's convenient. This is open weights because Met…

It’s rather amusing to me to read comments like this, and then simultaneously whenever a Chinese company or team releases open-weight models or whatever there is a giant round of applause, America is so behind, and there’s nothing but positive things to say about the intelligent, creative, and well-intentioned Chinese engineers (which is true, America certainly doesn’t have a monopoly on great people). Don’t you know…

Can you point to a lot of posts lauding the Chinese government based on the release of Chinese open models? Because that would be the equivalent to contrast with the OP.

Re: Muse Glimmer: 30B-parameter model optimized for always-on local agent workflows

#292

Post by bwfan123 »

solarkraft wrote: post #11

Wow, Meta is back (at least for now)! I like this class of model. Multi-token prediction makes it viable to run dense models at not-too-far-off speeds as MoE models with much better intelligence. The submission’s title (open weights 30B local coding model) is luckily wrong: This is meant to be a general agentic model. It even comes pre-quantized and with a MTP/drafter model. Looking good! Let’s hope they aren’t disho…

> It even comes pre-quantized and with a MTP/drafter model

Glad to see the extra engineering effort that went into creating this local model and making it run well on a consumer device. I use qwen3.5-coder, and am waiting to kick the tires on this one. I hate to say this, but kudos to Meta ! I hope apple and others follow suit and create similar local models for other use cases like audio, images and video that can run on a laptop.

Re: Muse Glimmer: 30B-parameter model optimized for always-on local agent workflows

#294

Post by modzu »

mmaunder wrote: post #249

Remember when we needed 200 servers for an enterprise website because Apache used one process or thread per connection - and Nginx collapsed that into a single box overnight? That moment for LLMs is near. It’s going to move us from the big iron era of AI to small portable brains. Nature has already proved it’s possible with 20 watts and very little heat generation. And I think the data center buildout will end in car…

brains do it with 20 watts because theyre analog. llms require massive amounts of power and this isnt changing any time soon without a breakthrough

Re: Muse Glimmer: 30B-parameter model optimized for always-on local agent workflows

#295

Post by NorwegianDude »

OsamaJaber wrote: post #112

The comparison set is Gemma4-31B and Qwen3.6-27B, not the current Qwen Fair on size, but the headline numbers are against a model a generation back

That is the most recent Qwen and Google models, there is no newer version, yet. Qwen3.8 27B might come in a couple of days tho, if it's launched alongside the large one when the Qwen3.8 countdown reaches zero.

Re: Muse Glimmer: 30B-parameter model optimized for always-on local agent workflows

#296

Post by stiltzkin »

ericmay wrote: post #171
avaer wrote: post #101

I lament the comments saying this in any way redeems Meta (the company). The researchers releasing this stuff have almost nothing to do with Meta other than being bankrolled by the slaughterhouse. You aren't the customer, you are the pawn in big tech's game of thrones. Your good will is a commodity to be traded, almost literally. It will be used against you the moment it's convenient. This is open weights because Met…

It’s rather amusing to me to read comments like this, and then simultaneously whenever a Chinese company or team releases open-weight models or whatever there is a giant round of applause, America is so behind, and there’s nothing but positive things to say about the intelligent, creative, and well-intentioned Chinese engineers (which is true, America certainly doesn’t have a monopoly on great people). Don’t you know…

[dead]

Re: Muse Glimmer: 30B-parameter model optimized for always-on local agent workflows

#297

Post by ericmay »

applfanboysbgon wrote: post #286
ericmay wrote: post #235

Earlier quoted context omitted.

Not applied accurately with respect to my comment, but it is a funny one and also new to me in the naming.

> It’s rather amusing to me to read comments like this, and then simultaneously whenever a Chinese company or team releases open-weight models or whatever there is a giant round of applause, America is so behind, and there’s nothing but positive things to say about the intelligent, creative, and well-intentioned Chinese engineers It is absolutely applied accurately. You're commenting on the alleged hypocrisy of peopl…

[flagged]

Re: Muse Glimmer: 30B-parameter model optimized for always-on local agent workflows

#298

Post by tgtweak »

androiddrew wrote: post #293

I'd really like to see a 45B-ish dense model ready for a dual GPU setup. Something with a little more intelligence while still within the range of some higher end local setups.

There is definitely an under-served target memory size of 48GB - almost everything aims for: 12, 16, 24, 32, 64, ...) But most dual-gpu setups, 3090/4090 (and some mac configs afaik) have 48GB, and most 64GB systems would do well with the extra 16gb of overhead saved. 48GB is also moderately common in PC memory configurations since 24gb DIMMs are a thing.

Re: Muse Glimmer: 30B-parameter model optimized for always-on local agent workflows

#299

Post by dofm »

modzu wrote: post #294
mmaunder wrote: post #249

Remember when we needed 200 servers for an enterprise website because Apache used one process or thread per connection - and Nginx collapsed that into a single box overnight? That moment for LLMs is near. It’s going to move us from the big iron era of AI to small portable brains. Nature has already proved it’s possible with 20 watts and very little heat generation. And I think the data center buildout will end in car…

brains do it with 20 watts because theyre analog. llms require massive amounts of power and this isnt changing any time soon without a breakthrough

And a breakthrough in hardware, specifically.

Re: Muse Glimmer: 30B-parameter model optimized for always-on local agent workflows

#300

Post by Aurornis »

fwipsy wrote: post #192
ericmay wrote: post #171

Earlier quoted context omitted.

It’s rather amusing to me to read comments like this, and then simultaneously whenever a Chinese company or team releases open-weight models or whatever there is a giant round of applause, America is so behind, and there’s nothing but positive things to say about the intelligent, creative, and well-intentioned Chinese engineers (which is true, America certainly doesn’t have a monopoly on great people). Don’t you know…

Good points, I personally believe that if/when China takes the lead, they will immediately stop releasing model weights. It only makes sense as a strategy to counterbalance (current) American labs' monopoly on frontier models. Holding both those positions would be hypocritical all right, but are you sure it's the same people commenting/voting in both cases? I don't think there's a strong consensus on Hacker News. Eve…

> I don't think there's a strong consensus on Hacker News.

There are diverse viewpoints. However there are some topics and threads where it becomes obvious that the comments are going to tilt toward one viewpoint. Participating in those threads with a different opinion will get your comments downvoted to -2 within minutes even if it’s well-written and factually sound.

After this happens a couple times you learn not to engage with those threads because it only takes a few zealous downvoters to bury anything you write. So the illusion of consensus persists.

Concrete example: There was that fake (AI hallucinated) report that Meta spent $2B lobbying on something that was popular here months ago. I actually read the repo and report and noticed the AI hallucination, as well as pointed out that $2B in lobbying spend by a single company was not plausible or supported by any evidence. It didn’t matter how I wrote it, it would risk getting downvotes and angry replies about “How dare you defend Meta!” Some people are here for the anger and to feel revenge against the enemies they think they know (like the US) and will cheer on anything that goes against those enemies, regardless of the other facts surrounding it. Factually accuracy often takes a back seat to pushing agendas.

Post reply on HN