Google DeepMind Cuts Over One-Third of Staff, Halts Pro Model Updates

DeepMind plans to dismiss 30%+ of its 7,000-8,000 workforce while suspending Pro model updates. Resources shift to cheaper Flash models amid low OKR scores and TPU constraints.

Woofun AI reports that Google DeepMind is executing a major restructuring plan involving the dismissal of more than one-third of its current workforce, which numbers between 7,000 and 8,000 employees. The initiative aims to streamline operations by removing redundant roles that do not align with actual tasks.

The company will temporarily suspend updates for its Pro model and reduce investment in large flagship models such as Fable and Opus. Instead, resources are being redirected toward the Flash model, which offers lower costs and higher update frequency. This strategic pivot coincides with a recent OKR score of 0.5 out of 1, reflecting difficulties in securing training resources for large-scale models due to high TPU demand from core services like Search, YouTube, and Gmail.