...how? I'd understand a 20-30% price drop from infra improvements for a model as-is, but 80%?
I wonder if "we quantized it lol" would classify as false advertising for modern LLMs.
Presumably because the price was about 5x higher to begin with than any the competitors at the same tier of performance? Perhaps it's better to get paid anything at all than to just lose 100% of the customers.
Deepseek made a few major innovations allowing them to achieve major compute efficiency and then published them. My guess is that OpenAI just implemented these themselves.
Wouldn’t surprise me. And even with this price cut it’s still 4x more expensive than Deepseek R1 is.