GPT-4 is getting cheaper and faster — but will it be enough to stop open source?
openai's new full-stack roadmap details how they plan to lower costs and increase performance across their platform.
openai outlined a full-stack strategy to build 'abundant intelligence' by optimizing ai from the hardware layer up. the plan includes developing custom silicon and rearchitecting software to significantly lower costs and boost model performance. this move signals a shift from relying solely on Nvidia and addresses the growing threat of efficient open-source models like Llama 3. openai aims to make gpt-4 class capabilities universally accessible.