Minimax-M3 running on 4x RTX Pro 6000s
- 800k context
- 4x concurrency at 250k
- 70-120 tok/s
- 2000 tok/s prefill no cache
- 376gb vram
- mxfp4
It's working on improving the audio on one of my videos, it's actually doing a good job in researching solutions.
Good model
This video is such a stretch and so over dramatized. Not sure what regulation they’re expecting or trying to push for. Why is there never any discussions about personal accountability?
https://t.co/wMvXosKei4
@realTimTraveler They add new features that haven’t been fully tested on scale. Bugs might exist in certain cases which have not shown up yet but will when used by millions of people
@AlexanderGlista@Metrolinx They could’ve spent a little extra during the build and saved millions that will be spent on maintenance over the years. Also the puddle on the tracks does not look safe. Are they not building a digital sim and stress testing before they build out these projects?
@Polymarket Digital ID is coming for everyone and every country. No way around it. People can complain but they’ll be the first to use it when it’s convenient
@Polymarket And this is why we can’t rely on the closed source model providers. Need open source models or we’re at the mercy of whoever has a model. Shout to China for actually putting out open source models. But we need more from every country
It’s opened so many new doors. For learners, it’s uncapped potential. Instead of one man becoming tony stark, basically everyone can now have their own Jarvis today if they care to set it up.
Also, one thing to note is that the models are building a database of every human. It will be able to predict the humans next move similar to minority report/westworld. Not sure if it’s a good or a bad thing in the future but right now there’s infinitely more benefits to having AI vs not. Reality seems to be much different than how it’s depicted in movies (The Jarvis example)
Hardware is getting better optimized for local inference. I think being able to train a local AI easily and cheaply is the next logical step. Rn we’re relying on the big providers. We’re definitely living through one of the greatest moments in history.
Can definitely pull way better returns elsewhere but $SPCX is a long term bet and they’ll definitely push past $10T in a decade. More of a mission driven company and it’ll be a safe bet for people to park their funds. They’re achieving things no other company can atm and at a speed where no one will realistically catch up for decades