Inkling: A New Open-Weight 975B MoE with a Few Surprises

Architecture and benchmark notes on Thinking Machines Lab's 975B Inkling MoE, including short convolutions, relative-position bias, training, and effort control.