This model is better than ChatGPT and 10x cheaper

[@pdmarino]

Thanks for the information. I uploaded the paper to NotebookLM and am querying it now. Waiting on the audio overview podcast to create.

[@johang1293]

Looking forward to seeing what model size they release next.

[@TimBrooks]

I cant figure out why the light switch keeps wiggling.

[@SK-gc7xv]

It looks like he's recording from a laptop that's on a stand on a couch, so as he moves, the couch cushion moves, making the laptop move a bit,

[@NateBJones]

yeah substandard recording options here on vacay

[@SK-gc7xv]

@@NateBJones Dammit, I want my money back. Oh...

[@MThomasB]

Thank you for the information Nate

[@aiforculture]

Thanks for this, really interesting. Will experiment with it! Mixture of expert style models definitely seem to have some brilliant applications.

[@giuseppevalenza3964]

Thanks for sharing over and over , great quality informative videos. Look forward to the next :)

[@AnonymousThinker-zz9dz]

A lot of startup have $5 million?! Damn even small business is big

[@bonecircuit]

Thanks for taking the time to make the video. Are you interested in the Sparse revolution by chance?

[@SK-gc7xv]

What about a model that is more efficient like this one, but then runs like o1 on that more efficient, focused way?

[@AK-vx4dy]

But what is special op ? Give model cheap to spread then it frees and spy or lockdown some countires ?

[@RasmusSchultz]

everyone keeps saying this is open source. where is the source? looks like open model only. 🤔

[@kaio0777]

nice that good data.

[@SK-gc7xv]

QwQ is pretty good, too.

[@toxtethogrady4610]

still not better than claude

[@abdusalamolamide]

coping

[@wwkk4964]

It has surpassed Claude for a while. Ask Claude any question that requires thinking like Maths and it will fail, try deepseek r1, it will crush. see the benchmarks yourself.