I think the exact numbers may be off but the important thing is the pricing model on the API being many times cheaper than others. I suspect deepseek is unable to lose the same amount of money as the infinitely funded blackholes that are Anthropic and OpenAI
Artificial Intelligence thread
- Thread starter 9dashline
- Start date
China Aluminium corp is incorporating 坤安 large model in its process. This is the first major model for non-ferrous metal industry.
not doubt. They and other Chinese AI players have to be more efficient with their computation resources for training. They are also essentially utilizing Western open source cloud platform for inference in platforms like Huggingface and such. Great strategy by the way.
It has been difficult to trust what comes out of OpenAI after the whole Sora stuff
paid... the newest version. it really is better than sora...
Deepseek definitely has its share of admirers in the West. This is from Andrej karpathy
I am still not convinced their model is really trained on that few GPUs. But it’s clear that they are very efficient.
So far, they are the story of china’s GenAI movement.
May least when it comes to non visual stuff.
I am still not convinced their model is really trained on that few GPUs. But it’s clear that they are very efficient.
So far, they are the story of china’s GenAI movement.
May least when it comes to non visual stuff.
It should be. Kuaishou has so much more and better data to train on than openAI.
I am still waiting on bytedance to come out with a game changing multi modal model. They have so much global data.
Yes, Veo 2 > Kling > Sora
It’s sad how China, dancing with export ban shackles, is able to beat the U.S. at pretty much anything nowadays. From EV/self driving to next gen fighter aircraft/EWAC, from video games to LLM/AI models. If they continue at this pace not even core tech like EUV and stepmom pron are safe.