alt Hacker NewsThis latest DeepSeek is almost at the "too cheap to meter" level. That's going to be a larger unlock than models like Fable/Mythos that are way too expensive to justify, IMO.
What secret sauce do they have?
Replies
throwaway_95283 • yesterday at 6:44 PM
limited resources, no modern GPUs, no $10 billion dev budgets.
pair it with codewhale, 50 agents, 200 MB of ram.
pama • yesterday at 7:59 PM
No secrets—all published. Very efficient attention. Excellent kernels. Great caching subsystem. Small and well trained model.
> What secret sauce do they have?
Quant company usually squeezing every penny.