📅 Nov 19, 2026
OpenAI says its models squeezed up to 14% more out of AMD's flagship GPU design.
LexxWire
The models rewrote circuit code and beat baselines AMD engineers had already tuned for years.
LexxWire
OpenAI says the GPU work packed about a year of design progress into less than five months.
LexxWire
On Jalapeño, its first custom inference chip, model changes cut area 13% in the vector unit and 10% in the matrix unit — verified on real silicon.
LexxWire
OpenAI and Synopsys are training GPT-Synopsys to run chip design software the way expert engineers do.
LexxWire
Thanks for reading!