OP Pohjola turns to quantum-inspired AI compression
The expanded research partnership runs on classical computers, not quantum processors.
Reading tools for text size, copying, printing and your reading list

OP Pohjola and QuTwo are expanding their joint research centre into AI model compression. QuTwo announced the work on November 16; OP Pohjola issued its release on November 17. The aim is to make open language models more economical to use across the financial group.[1][2]
The method combines conventional compression with a quantum-inspired precision-recovery layer. It runs on today’s classical computers. This is not a demonstration of a quantum processor outperforming ordinary hardware.[1]
The announcement does not provide an OP-specific benchmark, measured cost savings or a deployment timetable. QuTwo cites previously announced model-quality results, rather than new banking performance data. QARC, the partners’ research centre, was established in February 2026.[1]