DeepSeek V4 Flash 0731 Intelligence, Performance and Price Analysis

[flagged]

[flagged]

[dead]

Can we please stop bringing artificial benchmaxxing shit to the top of HN?

Daily reminder that none of these numbers are valid in a world where no one publishes the sampling settings used.

Daily reminder that improving your samplers from the garbage default top_p/top_k to min_p or subsequent methods dramatically improves the performance of these models, and makes most quantities like measured "verbosity" and subsequent calculations of "intelligence per token" meaningless

Daily reminder that no one, including within academic AI research, AI engineers, normies, etc takes LLM sampling seriously enough.

Does it already know the answer to what happen at Tiananmen Square? Or still avoiding it?

wont AI models want to make themselves more intelligent and efficient by downloading 'better ' models? If the current models can break into openAI and Hugging face, arent they already breaking into to closed source repos which isnt publicized (so as not to harm stock valuations)? I am looking forward to when these cyberweapons break loose. It will be like a software version of COVID. It will be wonderful when humans become valuable again.