The strategy helps companies avoid dependency on frontier AI labs
something has definitely shifted in the past few weeks. seeing a huge uptick in large enterprises wanting to secure compute and post-train their own models in house, frequently on top of GLM-5.2. everyone is starting to understand how open source wins.
Many users hail GLM-5.2 as a watershed enabling open models to replace closed ones for enterprise in-house use and fine-tuning, while others question the need for post-training or flag compute bottlenecks and supply risks.
No Digg Deeper questions have been answered for this story yet.
Is what happens when the world becomes AGI pilled then both the leading lab and the government tell you you need to bow down if you want access to their models.
I feel it too. More of the last few weeks giving people the words to explain how they felt for months.
something has definitely shifted in the past few weeks. seeing a huge uptick in large enterprises wanting to secure compute and post-train their own models in house, frequently on top of GLM-5.2. everyone is starting to understand how open source wins.
@willccbb tap the sign. a new era. https://www.interconnects.ai/p/glm-52-is-the-step-change-for-open
something has definitely shifted in the past few weeks. seeing a huge uptick in large enterprises wanting to secure compute and post-train their own models in house, frequently on top of GLM-5.2. everyone is starting to understand how open source wins.
@willccbb related
@willccbb Yup
if you wanna build the open infra of the future, come join us
@willccbb I'm not even sure most of them need post-training at this point
@willccbb what happened is they got their $100k anthropic/ai bill
@willccbb @yacineMTB we just needed fable at home
@willccbb Big plus one.
@willccbb Meh. Care more about the economics of open source more than post training
@willccbb yup confirm
@willccbb open models will soon be significantly caught up with the frontier closed model perf, but will fall short slightly cause of the closed loop compute and entrp support (not significant diff though in perf), its good to see open source winning again
@willccbb It’s happening
@willccbb gonna have to restart the RLM training run to be based on 5.2 now...
@willccbb and a sudden need of b300s
@willccbb anthropic overplayed it hand, and got regulated, now it is being viewed as a supply chain risk.
@natolambert That's why local models are so important. If you’re running inference on your own machine, no one can just shut you out or change the rules overnight.
@willccbb Nope 👎
@willccbb yes people want the models deployed in house on their own server
@willccbb The CFOs got the bill and API spend is now a large enough cost to be actively managed as part of typical financial management.